fix(mcp): make capability resolver side-effect free (Closes #685)

Stale-runtime detection remains fail-closed, but gitea_resolve_task_capability
must never touch mcp_config.json, spawn recovery threads, or call os._exit.

- Remove _trigger_mcp_auto_restart from the read-only diagnostics path
- Return blocker_kind=runtime_reconnect_required with mutation_performed=false
- Keep exact_safe_next_action pointing at IDE/client reconnect when stale
- Add regression suite across author/reviewer/merger/reconciler profiles

.env.example

@@ -55,3 +55,12 @@ GITEA_MCP_PROFILE=prgs
 # GITEA_MERGER_WORKTREE=/path/to/repo/branches/merge-pr456
 # GITEA_RECONCILER_WORKTREE=/path/to/repo/branches/reconcile-pr456
 # GITEA_ACTIVE_WORKTREE=/path/to/repo/branches/session-override
+
+# Durable HMAC key for irrecoverable decision-lock provenance artifacts (#709 F4).
+# REQUIRED in production native MCP processes that mint or verify recovery
+# authorization (reconciler + merger must share the same key). Hex (preferred),
+# base64, or utf-8 literal (>=16 bytes). Never generate an ephemeral per-process
+# key — cross-process / post-restart verification would fail closed.
+# GITEA_IRRECOVERABLE_AUTH_HMAC_KEY=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef
+# Optional key version string bound into the signature (for rotation).
+# GITEA_IRRECOVERABLE_AUTH_HMAC_KEY_VERSION=v1

README.md

@@ -53,6 +53,7 @@ Any MCP-compatible agent (Antigravity, Claude Code, etc.) can call these tools n
 | `gitea_whoami` | Read-only: identify the authenticated Gitea account (safe metadata only) |
 | `gitea_get_profile` | Read-only: describe the active runtime execution profile (safe metadata only) |
 | `gitea_check_pr_eligibility` | Read-only: check if the current identity/profile may review/approve/request_changes/merge a PR |
+| `gitea_assess_conflict_fix_classification` | Read-only: classify conflict-fix need from a live PR head re-fetch before creating a conflict-fix worktree |
 | `gitea_submit_pr_review` | Gated review mutation: comment/approve/request_changes, only after identity+profile+eligibility gates pass (no merge, no self-approval) |
 | `gitea_mark_issue` | Claim/release an issue (start/done) |
 | `gitea_list_labels` | List all available labels in a repository |

allocator_service.py

@@ -0,0 +1,610 @@
+"""Controller-owned work allocator policy (#600).
+
+Builds on the #613 control-plane DB substrate (``ControlPlaneDB.assign_and_lease``).
+
+Workers must not self-select exclusive work under the standard multi-LLM
+workflow. They call ``gitea_allocate_next_work`` which:
+
+1. Inspects candidate Gitea issues/PRs (never raw monitoring incidents).
+2. Applies ADR routing policy (role, terminal path, leases, blocked, deps).
+3. Atomically assigns + leases the selected item in one DB transaction.
+
+This module is pure selection + substrate orchestration. Gitea I/O for live
+inventory lives in the MCP tool wrapper so tests can inject candidates.
+
+#612 remains downstream: bridge-created Gitea issues become candidates only
+after they exist as normal issues; this module never assigns incidents.
+"""
+
+from __future__ import annotations
+
+import os
+import uuid
+from dataclasses import dataclass, field
+from typing import Any, Sequence
+
+from control_plane_db import (
+    ControlPlaneDB,
+    ControlPlaneError,
+    InvalidWorkKindError,
+    LeaseRequiredError,
+    WORK_KINDS,
+)
+
+# Outcomes required by #600 / ADR §5.
+OUTCOME_ASSIGNED = "assigned_work"
+OUTCOME_WAIT = "wait"
+OUTCOME_BLOCKED_TERMINAL = "blocked_by_terminal_path"
+OUTCOME_BLOCKED_LEASE = "blocked_by_active_lease"
+OUTCOME_NEEDS_CONTROLLER = "needs_controller"
+OUTCOME_NO_SAFE = "no_safe_work"
+OUTCOME_ROLE_INELIGIBLE = "role_ineligible"
+OUTCOME_PREVIEW = "preview"  # dry-run only (apply=false)
+
+ROLE_AUTHOR = "author"
+ROLE_REVIEWER = "reviewer"
+ROLE_MERGER = "merger"
+ROLE_RECONCILER = "reconciler"
+ROLE_CONTROLLER = "controller"
+
+VALID_ROLES = frozenset(
+    {ROLE_AUTHOR, ROLE_REVIEWER, ROLE_MERGER, ROLE_RECONCILER, ROLE_CONTROLLER}
+)
+
+# Default action matrices by role (mutation gate will re-check).
+ROLE_ACTIONS: dict[str, tuple[tuple[str, ...], tuple[str, ...]]] = {
+    ROLE_AUTHOR: (
+        ("implement", "comment", "push", "create_pr"),
+        ("approve", "merge", "request_changes", "self_select_without_assignment"),
+    ),
+    ROLE_REVIEWER: (
+        ("review", "comment", "approve", "request_changes"),
+        ("merge", "push", "create_pr", "self_select_without_assignment"),
+    ),
+    ROLE_MERGER: (
+        ("merge", "comment"),
+        ("approve", "request_changes", "push", "create_pr", "self_select_without_assignment"),
+    ),
+    ROLE_RECONCILER: (
+        ("comment", "diagnose", "cleanup"),
+        ("approve", "merge", "push", "create_pr", "self_select_without_assignment"),
+    ),
+    ROLE_CONTROLLER: (
+        ("comment", "diagnose", "allocate"),
+        ("approve", "merge", "push", "create_pr"),
+    ),
+}
+
+
+@dataclass
+class WorkCandidate:
+    """One assignable Gitea issue or PR presented to the allocator."""
+
+    kind: str  # issue | pr
+    number: int
+    state: str = "open"
+    labels: tuple[str, ...] = ()
+    title: str = ""
+    priority: int = 0
+    head_sha: str | None = None
+    # Routing signals (callers derive from Gitea / review feedback).
+    request_changes_current_head: bool = False
+    approval_on_current_head: bool = False
+    approval_stale: bool = False
+    approval_contaminated: bool = False
+    mergeable: bool = False
+    blocked: bool = False
+    dependency_unmet: bool = False
+    dependency_reason: str | None = None
+    already_claimed_elsewhere: bool = False
+
+    def __post_init__(self) -> None:
+        self.kind = (self.kind or "").strip().lower()
+        self.state = (self.state or "open").strip().lower()
+        self.labels = tuple(
+            str(x).strip().lower() for x in (self.labels or ()) if str(x).strip()
+        )
+        if self.kind not in WORK_KINDS:
+            raise InvalidWorkKindError(
+                f"candidate kind '{self.kind}' is not assignable; only "
+                f"{sorted(WORK_KINDS)} (never raw incidents)"
+            )
+
+    def as_dict(self) -> dict[str, Any]:
+        return {
+            "kind": self.kind,
+            "number": self.number,
+            "state": self.state,
+            "labels": list(self.labels),
+            "title": self.title,
+            "priority": self.priority,
+            "head_sha": self.head_sha,
+            "request_changes_current_head": self.request_changes_current_head,
+            "approval_on_current_head": self.approval_on_current_head,
+            "approval_stale": self.approval_stale,
+            "approval_contaminated": self.approval_contaminated,
+            "mergeable": self.mergeable,
+            "blocked": self.blocked,
+            "dependency_unmet": self.dependency_unmet,
+            "dependency_reason": self.dependency_reason,
+        }
+
+
+@dataclass
+class SkipRecord:
+    kind: str
+    number: int
+    reason: str
+
+    def as_dict(self) -> dict[str, Any]:
+        return {"kind": self.kind, "number": self.number, "reason": self.reason}
+
+
+def normalize_role(role: str | None, *, profile_name: str | None = None) -> str:
+    """Map profile/role strings to a canonical allocator role."""
+    raw = (role or "").strip().lower()
+    if raw in VALID_ROLES:
+        return raw
+    prof = (profile_name or "").strip().lower()
+    for token in VALID_ROLES:
+        if token in prof or prof.endswith(f"-{token}"):
+            return token
+    if "author" in raw:
+        return ROLE_AUTHOR
+    if "review" in raw:
+        return ROLE_REVIEWER
+    if "merg" in raw:
+        return ROLE_MERGER
+    if "reconcil" in raw:
+        return ROLE_RECONCILER
+    if "control" in raw:
+        return ROLE_CONTROLLER
+    raise ControlPlaneError(
+        f"unknown allocator role '{role}' (profile={profile_name!r}); "
+        f"expected one of {sorted(VALID_ROLES)}"
+    )
+
+
+def expected_role_for_candidate(c: WorkCandidate) -> str:
+    """ADR §5.3 routing: which role should take this work next."""
+    if c.kind == "pr":
+        if c.approval_contaminated:
+            return ROLE_RECONCILER
+        if c.request_changes_current_head:
+            return ROLE_AUTHOR
+        if c.approval_stale:
+            return ROLE_REVIEWER
+        if c.approval_on_current_head and c.mergeable:
+            return ROLE_MERGER
+        # Open PR without terminal verdict → reviewer
+        return ROLE_REVIEWER
+    # Issues: ready work → author by default; blocked stays controller/none
+    labels = set(c.labels)
+    if "status:blocked" in labels or c.blocked:
+        return ROLE_CONTROLLER
+    return ROLE_AUTHOR
+
+
+def role_actions(role: str) -> tuple[tuple[str, ...], tuple[str, ...]]:
+    return ROLE_ACTIONS.get(role, ROLE_ACTIONS[ROLE_AUTHOR])
+
+
+def classify_skip(
+    c: WorkCandidate,
+    *,
+    role: str,
+    terminal_pr: int | None,
+) -> str | None:
+    """Return skip reason, or None if candidate is selectable for *role*."""
+    if c.state in ("merged", "closed"):
+        return f"{c.kind}#{c.number} is {c.state}; never assign"
+    if c.blocked or "status:blocked" in c.labels:
+        return f"{c.kind}#{c.number} is blocked"
+    if c.dependency_unmet:
+        return (
+            c.dependency_reason
+            or f"{c.kind}#{c.number} has unmet dependencies"
+        )
+    if c.already_claimed_elsewhere:
+        return f"{c.kind}#{c.number} already claimed elsewhere"
+    if c.kind == "pr" and not (c.head_sha or "").strip():
+        return f"pr#{c.number} missing head_sha pin"
+
+    # Terminal path first: when an active terminal PR exists, only that PR
+    # (or controller diagnosis) is assignable for review-path roles.
+    if terminal_pr is not None and c.kind == "pr" and c.number != terminal_pr:
+        if role in (ROLE_REVIEWER, ROLE_MERGER):
+            return (
+                f"pr#{c.number} skipped: active terminal-review lock on "
+                f"PR #{terminal_pr} must be resolved first"
+            )
+
+    expected = expected_role_for_candidate(c)
+    if role == ROLE_CONTROLLER:
+        # Controller may inspect anything but only assigns diagnosis targets
+        # when contaminated / blocked.
+        if expected == ROLE_RECONCILER or c.blocked:
+            return None
+        return f"{c.kind}#{c.number} does not require controller (expected {expected})"
+
+    if role != expected:
+        return (
+            f"{c.kind}#{c.number} expects role '{expected}', active role is '{role}'"
+        )
+
+    # Ready-gate for issues: prefer status:ready when labels present.
+    if c.kind == "issue" and c.labels:
+        if "status:ready" not in c.labels and "status:in-progress" not in c.labels:
+            # Allow unlabeled open issues; only skip explicit non-ready states.
+            if any(l.startswith("status:") for l in c.labels):
+                return f"issue#{c.number} not status:ready ({','.join(c.labels)})"
+    return None
+
+
+def sort_candidates(candidates: Sequence[WorkCandidate]) -> list[WorkCandidate]:
+    """Higher priority first; then lower number (older issues) for stability."""
+    return sorted(
+        candidates,
+        key=lambda c: (-int(c.priority), c.kind != "pr", int(c.number)),
+    )
+
+
+def allocate_next_work(
+    db: ControlPlaneDB,
+    *,
+    session_id: str,
+    role: str,
+    remote: str,
+    org: str,
+    repo: str,
+    candidates: Sequence[WorkCandidate],
+    apply: bool = False,
+    profile_name: str | None = None,
+    username: str | None = None,
+    lease_ttl_seconds: int | None = None,
+) -> dict[str, Any]:
+    """Select and optionally reserve the next work unit via control-plane DB.
+
+    *apply=False* (default): dry-run selection only — no lease/assignment.
+    *apply=True*: atomic ``assign_and_lease`` for the selected candidate.
+
+    Never uses file locks or comment-only leases as the assignment source.
+    """
+    if db is None:
+        return {
+            "success": False,
+            "outcome": OUTCOME_NO_SAFE,
+            "reasons": [
+                "control-plane DB substrate unavailable (fail closed, #600/#613)"
+            ],
+            "skipped": [],
+            "assignment": None,
+            "substrate": "control_plane_db",
+            "file_lock_only": False,
+            "comment_lease_only": False,
+        }
+
+    try:
+        role_norm = normalize_role(role, profile_name=profile_name)
+    except ControlPlaneError as exc:
+        return {
+            "success": False,
+            "outcome": OUTCOME_ROLE_INELIGIBLE,
+            "reasons": [str(exc)],
+            "skipped": [],
+            "assignment": None,
+            "substrate": "control_plane_db",
+        }
+
+    session_id = (session_id or "").strip() or f"alloc-{uuid.uuid4().hex[:12]}"
+    try:
+        db.upsert_session(
+            session_id=session_id,
+            role=role_norm,
+            profile=profile_name,
+            pid=os.getpid(),
+        )
+    except Exception as exc:  # noqa: BLE001 — surface structured
+        return {
+            "success": False,
+            "outcome": OUTCOME_NO_SAFE,
+            "reasons": [
+                f"failed to register session in control-plane DB: {exc} "
+                "(fail closed, #613)"
+            ],
+            "skipped": [],
+            "assignment": None,
+            "substrate": "control_plane_db",
+        }
+
+    # Expire stale leases globally before selection.
+    try:
+        db.expire_stale_leases()
+    except Exception as exc:  # noqa: BLE001
+        return {
+            "success": False,
+            "outcome": OUTCOME_NO_SAFE,
+            "reasons": [f"lease expiry failed: {exc} (fail closed)"],
+            "skipped": [],
+            "assignment": None,
+            "substrate": "control_plane_db",
+        }
+
+    terminal = None
+    try:
+        terminal = db.get_active_terminal_lock(remote=remote, org=org, repo=repo)
+    except Exception as exc:  # noqa: BLE001
+        return {
+            "success": False,
+            "outcome": OUTCOME_NO_SAFE,
+            "reasons": [f"terminal lock lookup failed: {exc} (fail closed)"],
+            "skipped": [],
+            "assignment": None,
+            "substrate": "control_plane_db",
+        }
+    terminal_pr = int(terminal["terminal_pr"]) if terminal else None
+
+    skipped: list[SkipRecord] = []
+    ordered = sort_candidates(list(candidates))
+    selected: WorkCandidate | None = None
+    for c in ordered:
+        reason = classify_skip(c, role=role_norm, terminal_pr=terminal_pr)
+        if reason:
+            skipped.append(SkipRecord(c.kind, c.number, reason))
+            continue
+        selected = c
+        break
+
+    if selected is None:
+        # If terminal lock blocks all review work, surface that explicitly.
+        if terminal_pr is not None and role_norm in (ROLE_REVIEWER, ROLE_MERGER):
+            outcome = OUTCOME_BLOCKED_TERMINAL
+            reasons = [
+                f"no safe work for role '{role_norm}': active terminal-review "
+                f"lock on PR #{terminal_pr} (resolve terminal path first, #332/#600)"
+            ]
+        else:
+            outcome = OUTCOME_NO_SAFE
+            reasons = [
+                f"no safe assignable work for role '{role_norm}' "
+                f"among {len(ordered)} candidates"
+            ]
+        return {
+            "success": True,
+            "outcome": outcome,
+            "apply": bool(apply),
+            "role": role_norm,
+            "profile_name": profile_name,
+            "username": username,
+            "session_id": session_id,
+            "remote": remote,
+            "org": org,
+            "repo": repo,
+            "selected": None,
+            "expected_role_next": None,
+            "reasons": reasons,
+            "skipped": [s.as_dict() for s in skipped],
+            "terminal_pr": terminal_pr,
+            "assignment": None,
+            "substrate": "control_plane_db",
+            "file_lock_only": False,
+            "comment_lease_only": False,
+            "downstream_note": (
+                "#612 incident bridge remains downstream of #600; "
+                "allocator never assigns raw monitoring incidents"
+            ),
+        }
+
+    expected_role = expected_role_for_candidate(selected)
+    allowed, forbidden = role_actions(role_norm)
+    selection = {
+        "kind": selected.kind,
+        "number": selected.number,
+        "title": selected.title,
+        "labels": list(selected.labels),
+        "head_sha": selected.head_sha,
+        "priority": selected.priority,
+        "expected_role_next": expected_role,
+        "reason_selected": (
+            f"highest-priority candidate for role '{role_norm}' "
+            f"(expected_role={expected_role})"
+        ),
+    }
+
+    if not apply:
+        return {
+            "success": True,
+            "outcome": OUTCOME_PREVIEW,
+            "apply": False,
+            "role": role_norm,
+            "profile_name": profile_name,
+            "username": username,
+            "session_id": session_id,
+            "remote": remote,
+            "org": org,
+            "repo": repo,
+            "selected": selection,
+            "expected_role_next": expected_role,
+            "reasons": [
+                "dry-run only (apply=false); no assignment/lease created — "
+                "call again with apply=true to reserve via control-plane DB"
+            ],
+            "skipped": [s.as_dict() for s in skipped],
+            "terminal_pr": terminal_pr,
+            "assignment": None,
+            "substrate": "control_plane_db",
+            "file_lock_only": False,
+            "comment_lease_only": False,
+            "downstream_note": (
+                "#612 incident bridge remains downstream of #600; "
+                "allocator never assigns raw monitoring incidents"
+            ),
+        }
+
+    # Atomic reserve via #613 substrate.
+    ttl = lease_ttl_seconds if lease_ttl_seconds is not None else None
+    try:
+        kwargs: dict[str, Any] = {
+            "session_id": session_id,
+            "role": role_norm,
+            "remote": remote,
+            "org": org,
+            "repo": repo,
+            "kind": selected.kind,
+            "number": selected.number,
+            "expected_head_sha": selected.head_sha,
+            "allowed_actions": allowed,
+            "forbidden_actions": forbidden,
+            "phase": "allocated",
+        }
+        if ttl is not None:
+            kwargs["lease_ttl_seconds"] = int(ttl)
+        result = db.assign_and_lease(**kwargs)
+    except (InvalidWorkKindError, LeaseRequiredError, ControlPlaneError) as exc:
+        return {
+            "success": False,
+            "outcome": OUTCOME_NO_SAFE,
+            "apply": True,
+            "role": role_norm,
+            "session_id": session_id,
+            "remote": remote,
+            "org": org,
+            "repo": repo,
+            "selected": selection,
+            "expected_role_next": expected_role,
+            "reasons": [f"atomic assign+lease failed: {exc} (fail closed, #613)"],
+            "skipped": [s.as_dict() for s in skipped],
+            "terminal_pr": terminal_pr,
+            "assignment": None,
+            "substrate": "control_plane_db",
+            "file_lock_only": False,
+            "comment_lease_only": False,
+        }
+
+    if result.outcome == "wait":
+        return {
+            "success": True,
+            "outcome": OUTCOME_WAIT,
+            "apply": True,
+            "role": role_norm,
+            "session_id": session_id,
+            "remote": remote,
+            "org": org,
+            "repo": repo,
+            "selected": selection,
+            "expected_role_next": expected_role,
+            "reasons": [result.reason or "foreign active lease"],
+            "skipped": [s.as_dict() for s in skipped],
+            "terminal_pr": terminal_pr,
+            "assignment": result.as_dict(),
+            "owner_session_id": result.owner_session_id,
+            "substrate": "control_plane_db",
+            "file_lock_only": False,
+            "comment_lease_only": False,
+        }
+
+    if result.outcome == "no_safe_work":
+        return {
+            "success": True,
+            "outcome": OUTCOME_NO_SAFE,
+            "apply": True,
+            "role": role_norm,
+            "session_id": session_id,
+            "remote": remote,
+            "org": org,
+            "repo": repo,
+            "selected": selection,
+            "expected_role_next": expected_role,
+            "reasons": [result.reason or "no_safe_work"],
+            "skipped": [s.as_dict() for s in skipped],
+            "terminal_pr": terminal_pr,
+            "assignment": result.as_dict(),
+            "substrate": "control_plane_db",
+            "file_lock_only": False,
+            "comment_lease_only": False,
+        }
+
+    # assigned
+    return {
+        "success": True,
+        "outcome": OUTCOME_ASSIGNED,
+        "apply": True,
+        "role": role_norm,
+        "profile_name": profile_name,
+        "username": username,
+        "session_id": session_id,
+        "remote": remote,
+        "org": org,
+        "repo": repo,
+        "selected": selection,
+        "expected_role_next": expected_role,
+        "reasons": [
+            selection["reason_selected"],
+            result.reason or "atomic assign+lease created",
+        ],
+        "skipped": [s.as_dict() for s in skipped],
+        "terminal_pr": terminal_pr,
+        "assignment": result.as_dict(),
+        "lease_proof": {
+            "assignment_id": result.assignment_id,
+            "lease_id": result.lease_id,
+            "expires_at": result.expires_at,
+            "expected_head_sha": result.expected_head_sha,
+            "allowed_actions": list(result.allowed_actions),
+            "forbidden_actions": list(result.forbidden_actions),
+            "source": "control_plane_db.assign_and_lease",
+        },
+        "next_valid_command": _next_command(role_norm, selected),
+        "substrate": "control_plane_db",
+        "file_lock_only": False,
+        "comment_lease_only": False,
+        "downstream_note": (
+            "#612 incident bridge remains downstream of #600; "
+            "allocator never assigns raw monitoring incidents"
+        ),
+    }
+
+
+def _next_command(role: str, c: WorkCandidate) -> str:
+    if role == ROLE_AUTHOR and c.kind == "issue":
+        return f"implement issue #{c.number} under a branches/ worktree; open PR when ready"
+    if role == ROLE_AUTHOR and c.kind == "pr":
+        return (
+            f"address REQUEST_CHANGES on PR #{c.number} at head "
+            f"{(c.head_sha or '')[:12]} and push fixes"
+        )
+    if role == ROLE_REVIEWER:
+        return (
+            f"review PR #{c.number} pinned at head {(c.head_sha or '')[:12]} "
+            "via full reviewer workflow"
+        )
+    if role == ROLE_MERGER:
+        return (
+            f"merge PR #{c.number} only with explicit operator MERGE "
+            f"authorization at head {(c.head_sha or '')[:12]}"
+        )
+    if role == ROLE_RECONCILER:
+        return f"diagnose contested state for {c.kind}#{c.number}"
+    return f"proceed on {c.kind}#{c.number} under role {role}"
+
+
+def candidate_from_dict(data: dict[str, Any]) -> WorkCandidate:
+    """Build a WorkCandidate from a plain dict (tests / MCP inventory)."""
+    return WorkCandidate(
+        kind=str(data.get("kind") or "issue"),
+        number=int(data["number"]),
+        state=str(data.get("state") or "open"),
+        labels=tuple(data.get("labels") or ()),
+        title=str(data.get("title") or ""),
+        priority=int(data.get("priority") or 0),
+        head_sha=data.get("head_sha"),
+        request_changes_current_head=bool(data.get("request_changes_current_head")),
+        approval_on_current_head=bool(data.get("approval_on_current_head")),
+        approval_stale=bool(data.get("approval_stale")),
+        approval_contaminated=bool(data.get("approval_contaminated")),
+        mergeable=bool(data.get("mergeable")),
+        blocked=bool(data.get("blocked")),
+        dependency_unmet=bool(data.get("dependency_unmet")),
+        dependency_reason=data.get("dependency_reason"),
+        already_claimed_elsewhere=bool(data.get("already_claimed_elsewhere")),
+    )

branch_cleanup_guard.py

@@ -0,0 +1,603 @@
+"""Guards for merged-PR branch cleanup and raw git delete bypasses (#514)."""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+PROTECTED_BRANCHES = frozenset({"master", "main", "dev"})
+
+# Evidence / preservation branches must never be removed by cleanup tools
+# (e.g. chore/issue-681-preserve-review-session-wip).
+_PRESERVATION_MARKERS = ("preserve", "preservation", "evidence")
+
+
+def is_preservation_or_evidence_branch(branch: str | None) -> bool:
+    """Return True when *branch* is a preservation/evidence ref that must stay."""
+    if not branch:
+        return False
+    name = str(branch).lower()
+    return any(marker in name for marker in _PRESERVATION_MARKERS)
+
+
+_RAW_BRANCH_DELETE_PATTERNS = (
+    re.compile(r"\bgit(?:\s+-C\s+\S+)?\s+branch\s+-[dD]\b[^\n\r]*", re.I),
+    re.compile(r"\bgit(?:\s+-C\s+\S+)?\s+push\b[^\n\r]*\s--delete\b[^\n\r]*", re.I),
+    re.compile(r"\bgit(?:\s+-C\s+\S+)?\s+push\b[^\n\r]*\s:[^\s`]+", re.I),
+)
+
+
+def raw_branch_delete_commands(text: str | None) -> list[str]:
+    """Return raw git branch-delete commands cited in *text*."""
+    if not text:
+        return []
+    commands: list[str] = []
+    for pattern in _RAW_BRANCH_DELETE_PATTERNS:
+        commands.extend(match.group(0).strip("` ") for match in pattern.finditer(text))
+    return list(dict.fromkeys(commands))
+
+
+def assess_raw_branch_delete_report(text: str | None) -> dict[str, Any]:
+    """Fail closed when a report uses raw git branch deletion as cleanup proof."""
+    commands = raw_branch_delete_commands(text)
+    reasons = [
+        (
+            "raw git branch deletion bypasses MCP branch.delete cleanup gates: "
+            f"{command}"
+        )
+        for command in commands
+    ]
+    return {
+        "proven": not reasons,
+        "block": bool(reasons),
+        "commands": commands,
+        "reasons": reasons,
+        "safe_next_action": (
+            "use gitea_cleanup_merged_pr_branch or another approved cleanup "
+            "helper with explicit branch.delete capability"
+            if reasons
+            else "proceed"
+        ),
+    }
+
+
+def assess_merged_pr_branch_cleanup(
+    *,
+    pr_number: int,
+    head_branch: str,
+    merged: bool,
+    remote_branch_exists: bool,
+    open_pr_heads: set[str],
+    head_on_target: bool | None,
+    delete_capability_allowed: bool,
+    confirmation: str | None,
+    protected_branches: frozenset[str] | None = None,
+) -> dict[str, Any]:
+    """Assess whether a merged PR source branch can be deleted via MCP."""
+    protected = protected_branches or PROTECTED_BRANCHES
+    expected_confirmation = f"CLEANUP MERGED PR {pr_number} BRANCH {head_branch}"
+    reasons: list[str] = []
+    if not merged:
+        reasons.append("PR is not merged")
+    if not remote_branch_exists:
+        reasons.append("remote branch already absent")
+    if not head_branch:
+        reasons.append("PR head branch is missing")
+    if head_branch in protected:
+        reasons.append(f"branch '{head_branch}' is protected")
+    if is_preservation_or_evidence_branch(head_branch):
+        reasons.append(
+            f"branch '{head_branch}' is a preservation/evidence branch and "
+            "cannot be deleted through merged-PR cleanup"
+        )
+    if head_branch in open_pr_heads:
+        reasons.append("an open PR still references this head branch")
+    if head_on_target is False:
+        reasons.append("PR head is not an ancestor of the target branch")
+    if head_on_target is None:
+        reasons.append("PR head ancestry could not be proven")
+    if not delete_capability_allowed:
+        reasons.append("gitea.branch.delete capability is not allowed")
+    if confirmation != expected_confirmation:
+        reasons.append(
+            "confirmation must equal "
+            f"'{expected_confirmation}' for branch cleanup"
+        )
+
+    safe = not reasons
+    return {
+        "pr_number": pr_number,
+        "head_branch": head_branch,
+        "expected_confirmation": expected_confirmation,
+        "remote_branch_exists": remote_branch_exists,
+        "safe_to_delete": safe,
+        "block_reasons": reasons,
+        "recommended_action": "delete_remote_branch" if safe else "keep_remote_branch",
+    }
+
+
+# ---------------------------------------------------------------------------
+# #687 remediation: post-delete readback + active ownership protection
+# ---------------------------------------------------------------------------
+
+READBACK_NOT_FOUND = "not_found"
+READBACK_EXISTS = "exists"
+READBACK_AUTHENTICATION = "authentication_error"
+READBACK_AUTHORIZATION = "authorization_error"
+READBACK_TRANSPORT = "transport_error"
+READBACK_UNEXPECTED = "unexpected_response"
+READBACK_AMBIGUOUS_404 = "ambiguous_not_found"
+
+# Scope of a 404: only branch-scoped absence may set verified_absent.
+NOT_FOUND_SCOPE_BRANCH = "branch"
+NOT_FOUND_SCOPE_REPOSITORY = "repository"
+NOT_FOUND_SCOPE_HOST = "host"
+NOT_FOUND_SCOPE_UNKNOWN = "unknown"
+
+ERROR_CLASS_AUTHENTICATION = "authentication"
+ERROR_CLASS_AUTHORIZATION = "authorization"
+ERROR_CLASS_TRANSPORT = "transport"
+ERROR_CLASS_UNEXPECTED = "unexpected"
+
+OWNERSHIP_CATEGORY_AUTHOR_SESSION = "author_session"
+OWNERSHIP_CATEGORY_AUTHOR_LEASE = "author_lease"
+OWNERSHIP_CATEGORY_REVIEWER_LEASE = "reviewer_lease"
+OWNERSHIP_CATEGORY_MERGER_LEASE = "merger_lease"
+OWNERSHIP_CATEGORY_CONTROLLER_LEASE = "controller_lease"
+OWNERSHIP_CATEGORY_RECONCILER_LEASE = "reconciler_lease"
+OWNERSHIP_CATEGORY_WORKTREE_BINDING = "worktree_binding"
+OWNERSHIP_CATEGORY_INVENTORY_ERROR = "ownership_inventory_error"
+
+_ROLE_TO_OWNERSHIP_CATEGORY = {
+    "author": OWNERSHIP_CATEGORY_AUTHOR_LEASE,
+    "reviewer": OWNERSHIP_CATEGORY_REVIEWER_LEASE,
+    "merger": OWNERSHIP_CATEGORY_MERGER_LEASE,
+    "controller": OWNERSHIP_CATEGORY_CONTROLLER_LEASE,
+    "reconciler": OWNERSHIP_CATEGORY_RECONCILER_LEASE,
+}
+
+_ACTIVE_OWNERSHIP_STATUSES = frozenset(
+    {"active", "live", "claimed", "in_progress", "working", "pushing", "pushed"}
+)
+_TERMINAL_OWNERSHIP_STATUSES = frozenset(
+    {"released", "abandoned", "done", "blocked", "terminal", "closed"}
+)
+_EXPIRED_STATUSES = frozenset({"expired"})
+_STALE_STATUSES = frozenset({"stale", "stale_dead_process", "stale_missing_worktree"})
+
+
+def _norm_str(value: Any) -> str:
+    return str(value or "").strip()
+
+
+def normalize_host(host: str | None) -> str:
+    """Normalize a host identity for ownership matching (no credentials)."""
+    text = _norm_str(host).lower()
+    for prefix in ("https://", "http://"):
+        if text.startswith(prefix):
+            text = text[len(prefix) :]
+    # Drop path/query if a full URL slipped through.
+    text = text.split("/", 1)[0]
+    text = text.split("?", 1)[0]
+    return text.rstrip(".")
+
+
+def ownership_category_for_role(role: str | None) -> str:
+    """Map a role kind to a non-secret ownership category label."""
+    key = _norm_str(role).lower()
+    return _ROLE_TO_OWNERSHIP_CATEGORY.get(key, f"{key or 'unknown'}_lease")
+
+
+def classify_branch_readback_http_status(
+    status_code: int | None,
+    *,
+    not_found_scope: str | None = None,
+) -> dict[str, Any]:
+    """Classify a GET-branch HTTP status into a secret-free readback result.
+
+    R1: A bare/generic/repository/wrong-host 404 never yields
+    ``verified_absent=True``. Only an authoritative *branch-scoped* not-found
+    (``not_found_scope='branch'``) may verify deletion.
+    """
+    if status_code == 404:
+        scope = _norm_str(not_found_scope).lower() or NOT_FOUND_SCOPE_UNKNOWN
+        if scope == NOT_FOUND_SCOPE_BRANCH:
+            return {
+                "status": READBACK_NOT_FOUND,
+                "error_class": None,
+                "verified_absent": True,
+                "branch_present": False,
+                "not_found_scope": NOT_FOUND_SCOPE_BRANCH,
+                "reasons": [],
+            }
+        # repository / host / unknown / generic 404 — not verified absence
+        reason = {
+            NOT_FOUND_SCOPE_REPOSITORY: (
+                "post-delete readback 404 is repository-scoped, not branch absence"
+            ),
+            NOT_FOUND_SCOPE_HOST: (
+                "post-delete readback 404 is host-scoped, not branch absence"
+            ),
+        }.get(
+            scope,
+            "post-delete readback 404 is ambiguous (not branch-scoped); "
+            "cannot verify absence",
+        )
+        return {
+            "status": READBACK_AMBIGUOUS_404,
+            "error_class": ERROR_CLASS_UNEXPECTED,
+            "verified_absent": False,
+            "branch_present": None,
+            "not_found_scope": scope,
+            "reasons": [reason],
+        }
+    if status_code in (401, 407):
+        return {
+            "status": READBACK_AUTHENTICATION,
+            "error_class": ERROR_CLASS_AUTHENTICATION,
+            "verified_absent": False,
+            "branch_present": None,
+            "reasons": ["post-delete branch readback authentication failed"],
+        }
+    if status_code == 403:
+        return {
+            "status": READBACK_AUTHORIZATION,
+            "error_class": ERROR_CLASS_AUTHORIZATION,
+            "verified_absent": False,
+            "branch_present": None,
+            "reasons": ["post-delete branch readback authorization failed"],
+        }
+    if status_code is not None and 200 <= int(status_code) < 300:
+        return {
+            "status": READBACK_EXISTS,
+            "error_class": None,
+            "verified_absent": False,
+            "branch_present": True,
+            "reasons": ["post-delete readback found branch still present"],
+        }
+    if status_code is not None and int(status_code) >= 500:
+        return {
+            "status": READBACK_TRANSPORT,
+            "error_class": ERROR_CLASS_TRANSPORT,
+            "verified_absent": False,
+            "branch_present": None,
+            "reasons": ["post-delete branch readback transport/upstream failure"],
+        }
+    return {
+        "status": READBACK_UNEXPECTED,
+        "error_class": ERROR_CLASS_UNEXPECTED,
+        "verified_absent": False,
+        "branch_present": None,
+        "reasons": ["post-delete branch readback returned an unexpected response"],
+        "http_status": status_code,
+    }
+
+
+def _extract_http_status(exc: BaseException) -> int | None:
+    """Extract an HTTP status code from an exception chain (secret-free)."""
+    seen: set[int] = set()
+    current: BaseException | None = exc
+    while current is not None and id(current) not in seen:
+        seen.add(id(current))
+        for attr in ("code", "status", "status_code"):
+            value = getattr(current, attr, None)
+            if isinstance(value, int) and 100 <= value <= 599:
+                return value
+        text = str(current) if current is not None else ""
+        match = re.match(r"HTTP\s+(\d{3})\b", text)
+        if match:
+            return int(match.group(1))
+        current = current.__cause__ or current.__context__
+    return None
+
+
+def classify_branch_readback_exception(
+    exc: BaseException,
+    *,
+    not_found_scope: str | None = None,
+) -> dict[str, Any]:
+    """Classify a GET-branch exception without leaking credentials or bodies.
+
+    R1: substring ``404`` / ``not found`` alone never becomes verified_absent.
+    Callers must pass ``not_found_scope='branch'`` only after authoritative
+    proof that the repository/host is still reachable and the 404 is branch-level.
+    """
+    status_code = _extract_http_status(exc)
+    if status_code is None:
+        lower = str(exc).lower() if exc is not None else ""
+        if any(
+            token in lower
+            for token in (
+                "timed out",
+                "timeout",
+                "connection",
+                "network",
+                "temporarily unavailable",
+                "name or service not known",
+                "nodename nor servname",
+            )
+        ):
+            return {
+                "status": READBACK_TRANSPORT,
+                "error_class": ERROR_CLASS_TRANSPORT,
+                "verified_absent": False,
+                "branch_present": None,
+                "reasons": [
+                    "post-delete branch readback transport/upstream failure"
+                ],
+            }
+        if "unauthorized" in lower:
+            status_code = 401
+        elif "forbidden" in lower:
+            status_code = 403
+        elif "404" in lower or "not found" in lower:
+            # Ambiguous: do NOT treat as branch absence without scope proof.
+            status_code = 404
+            if not_found_scope is None:
+                not_found_scope = NOT_FOUND_SCOPE_UNKNOWN
+
+    if status_code is not None:
+        return classify_branch_readback_http_status(
+            status_code, not_found_scope=not_found_scope
+        )
+
+    return {
+        "status": READBACK_UNEXPECTED,
+        "error_class": ERROR_CLASS_UNEXPECTED,
+        "verified_absent": False,
+        "branch_present": None,
+        "reasons": ["post-delete branch readback returned an unexpected response"],
+    }
+
+
+def assess_post_delete_readback(readback: dict[str, Any] | None) -> dict[str, Any]:
+    """Decide whether DELETE success may be reported after branch readback.
+
+    Success requires authoritative branch-scoped not-found
+    (``verified_absent=True``). DELETE HTTP success alone is never enough.
+    Generic/repository/host 404 cannot verify absence (R1).
+    """
+    payload = dict(readback or {})
+    status = _norm_str(payload.get("status")) or READBACK_UNEXPECTED
+    # Only explicit verified_absent flag counts — never infer from status alone
+    # when status is a bare not_found without branch scope proof.
+    verified = bool(payload.get("verified_absent")) is True
+    if verified and status == READBACK_NOT_FOUND:
+        return {
+            "ok": True,
+            "success": True,
+            "verified_absent": True,
+            "readback": {
+                "status": READBACK_NOT_FOUND,
+                "verified_absent": True,
+                "branch_present": False,
+                "error_class": None,
+                "not_found_scope": NOT_FOUND_SCOPE_BRANCH,
+            },
+            "reasons": [],
+        }
+
+    reasons = list(payload.get("reasons") or [])
+    if not reasons:
+        if status == READBACK_EXISTS:
+            reasons = ["post-delete readback found branch still present"]
+        elif status == READBACK_AUTHENTICATION:
+            reasons = ["post-delete branch readback authentication failed"]
+        elif status == READBACK_AUTHORIZATION:
+            reasons = ["post-delete branch readback authorization failed"]
+        elif status == READBACK_TRANSPORT:
+            reasons = ["post-delete branch readback transport/upstream failure"]
+        elif status == READBACK_AMBIGUOUS_404:
+            reasons = [
+                "post-delete readback 404 is not branch-scoped; "
+                "cannot verify absence"
+            ]
+        else:
+            reasons = ["post-delete branch readback could not verify deletion"]
+
+    return {
+        "ok": False,
+        "success": False,
+        "verified_absent": False,
+        "readback": {
+            "status": status,
+            "verified_absent": False,
+            "branch_present": payload.get("branch_present"),
+            "error_class": payload.get("error_class"),
+            "not_found_scope": payload.get("not_found_scope"),
+        },
+        "reasons": reasons,
+        "blocker_kind": "post_delete_readback_failed",
+    }
+
+
+def cleanup_result_envelope(
+    *,
+    success: bool,
+    performed: bool,
+    delete_acknowledged: bool,
+    verified_absent: bool,
+    **extra: Any,
+) -> dict[str, Any]:
+    """R2: consistent top-level cleanup result fields on every return path."""
+    out: dict[str, Any] = {
+        "success": bool(success),
+        "performed": bool(performed),
+        "delete_acknowledged": bool(delete_acknowledged),
+        "verified_absent": bool(verified_absent),
+    }
+    out.update(extra)
+    return out
+
+
+def _repo_matches(
+    record: dict[str, Any],
+    *,
+    remote: str,
+    org: str,
+    repo: str,
+    host: str | None = None,
+) -> bool:
+    """Match ownership record to target remote/org/repo and normalized host."""
+    if (
+        _norm_str(record.get("remote")).lower() != _norm_str(remote).lower()
+        or _norm_str(record.get("org")).lower() != _norm_str(org).lower()
+        or _norm_str(record.get("repo")).lower() != _norm_str(repo).lower()
+    ):
+        return False
+    expected_host = normalize_host(host)
+    record_host = normalize_host(record.get("host") or record.get("host_name"))
+    # When both sides declare a host, they must agree after normalization.
+    # A record host that disagrees with the expected host is out of scope.
+    # Legacy records without host still match when remote/org/repo agree.
+    if expected_host and record_host and expected_host != record_host:
+        return False
+    return True
+
+
+def _branch_matches(record: dict[str, Any], branch: str) -> bool:
+    return _norm_str(record.get("branch")) == _norm_str(branch)
+
+
+def assess_ownership_record_activity(record: dict[str, Any]) -> dict[str, Any]:
+    """Classify one ownership record as blocking or non-blocking.
+
+    Distinguishes active ownership from expired/released/terminal/stale records.
+    Expired/stale records block unless reclaim_allowed is *explicitly* True
+    (O2: never treat missing/unknown reclaim as auto-allowed).
+    """
+    status = _norm_str(record.get("status")).lower()
+    category = _norm_str(record.get("category")) or "unknown"
+    reclaim_allowed = record.get("reclaim_allowed")
+
+    if category == OWNERSHIP_CATEGORY_INVENTORY_ERROR:
+        return {
+            "blocks": True,
+            "status": status or "unknown",
+            "category": category,
+            "reason": "ownership inventory failed closed",
+        }
+
+    if status in _TERMINAL_OWNERSHIP_STATUSES:
+        return {
+            "blocks": False,
+            "status": status,
+            "category": category,
+            "reason": f"{category} ownership is terminal/released ({status})",
+        }
+    if status in _ACTIVE_OWNERSHIP_STATUSES:
+        return {
+            "blocks": True,
+            "status": status,
+            "category": category,
+            "reason": f"active {category} ownership still uses the target branch",
+        }
+    if status in _EXPIRED_STATUSES or status in _STALE_STATUSES:
+        # O2: only explicit reclaim_allowed=True skips the block.
+        if reclaim_allowed is True:
+            return {
+                "blocks": False,
+                "status": status,
+                "category": category,
+                "reason": (
+                    f"{category} ownership is {status} and reclaimable; "
+                    "does not block deletion"
+                ),
+            }
+        return {
+            "blocks": True,
+            "status": status,
+            "category": category,
+            "reason": (
+                f"{status} {category} ownership still protects the target "
+                "branch (reclaim not proven; fail closed)"
+            ),
+        }
+    # Unknown status → fail closed
+    return {
+        "blocks": True,
+        "status": status or "unknown",
+        "category": category,
+        "reason": (
+            f"unclassified {category} ownership status "
+            f"'{status or 'unknown'}'; fail closed"
+        ),
+    }
+
+
+def assess_active_branch_ownership(
+    *,
+    remote: str,
+    org: str,
+    repo: str,
+    branch: str,
+    host: str | None = None,
+    records: list[dict[str, Any]] | None = None,
+) -> dict[str, Any]:
+    """Assess whether any active ownership still uses *branch* in *repo*.
+
+    Matching requires remote/org/repo/branch and, when provided, normalized
+    host identity. Other repositories, hosts, or branches never false-block.
+    """
+    target_branch = _norm_str(branch)
+    expected_host = normalize_host(host)
+    considered: list[dict[str, Any]] = []
+    blocking: list[dict[str, Any]] = []
+    ignored: list[dict[str, Any]] = []
+
+    for raw in records or []:
+        if not isinstance(raw, dict):
+            continue
+        if not _repo_matches(
+            raw, remote=remote, org=org, repo=repo, host=expected_host or None
+        ):
+            ignored.append(
+                {
+                    "category": _norm_str(raw.get("category")) or "unknown",
+                    "reason": "different repository or host scope",
+                }
+            )
+            continue
+        if not _branch_matches(raw, target_branch):
+            ignored.append(
+                {
+                    "category": _norm_str(raw.get("category")) or "unknown",
+                    "reason": "different branch",
+                }
+            )
+            continue
+        activity = assess_ownership_record_activity(raw)
+        entry = {
+            "category": activity["category"],
+            "status": activity["status"],
+            "blocks": activity["blocks"],
+            "reason": activity["reason"],
+        }
+        considered.append(entry)
+        if activity["blocks"]:
+            blocking.append(entry)
+
+    block = bool(blocking)
+    categories = sorted({b["category"] for b in blocking})
+    reasons = [
+        (
+            "active ownership protects branch "
+            f"'{target_branch}': " + "; ".join(b["reason"] for b in blocking)
+        )
+    ] if block else []
+    return {
+        "block": block,
+        "safe_to_delete": not block,
+        "remote": remote,
+        "org": org,
+        "repo": repo,
+        "host": expected_host or None,
+        "branch": target_branch,
+        "blocking_categories": categories,
+        "blocking": blocking,
+        "considered": considered,
+        "ignored_out_of_scope": ignored,
+        "reasons": reasons,
+        "blocker_kind": "active_branch_ownership" if block else None,
+        "recommended_action": "keep_remote_branch" if block else "delete_remote_branch",
+    }

canonical_comment_validator.py

@@ -0,0 +1,429 @@
+"""Fail-closed validation for workflow-changing Gitea comments (#496)."""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+VALID_ROLES = frozenset({
+    "controller",
+    "author",
+    "reviewer",
+    "merger",
+    "reconciler",
+    "user",
+})
+
+_BASE_REQUIRED = ("STATE", "WHO_IS_NEXT", "NEXT_ACTION", "NEXT_PROMPT", "WHY")
+
+_ISSUE_REQUIRED = _BASE_REQUIRED + ("BLOCKERS", "VALIDATION")
+_PR_REQUIRED = _BASE_REQUIRED + (
+    "ISSUE",
+    "HEAD_SHA",
+    "REVIEW_STATUS",
+    "MERGE_READY",
+    "BLOCKERS",
+    "VALIDATION",
+)
+_SUPERSESSION_REQUIRED = _BASE_REQUIRED + (
+    "CANONICAL_ITEM",
+    "SUPERSEDED_ITEM",
+    "CLOSE_OR_KEEP_OPEN",
+)
+
+_MACHINE_MARKERS = (
+    "<!-- mcp-review-lease:v1 -->",
+    "<!-- mcp-conflict-fix-lease:v1 -->",
+    "<!-- gitea-issue-claim-heartbeat:v1 -->",
+)
+
+_WORKFLOW_TRIGGERS = re.compile(
+    r"\b(?:"
+    r"blocked|unblocked|ready(?:\s+for\s+(?:review|merge|author))?|"
+    r"ready-to-merge|approved|approve|request\s+changes|changes\s+requested|"
+    r"superseded|duplicate|canonical|next\s+action|next\s+actor|who\s+is\s+next|"
+    r"author\s+should|reviewer\s+should|merger\s+should|reconciler\s+should|"
+    r"controller\s+should|issue\s+complete|pr\s+open|pr\s+merged|close\s+this|"
+    r"do\s+not\s+merge|needs?\s+rebase|stale\s+approval|contaminated\s+review|"
+    r"merge\s+ready|ready\s+for\s+merge"
+    r")\b",
+    re.IGNORECASE,
+)
+
+_CANONICAL_HEADINGS = (
+    "## Canonical Issue State",
+    "## Canonical PR State",
+    "## Canonical Discussion Summary",
+)
+
+_FIELD_RE = re.compile(
+    r"^([A-Z][A-Z0-9_]*)\s*:\s*(.*)$",
+    re.MULTILINE,
+)
+
+_VAGUE_NEXT_ACTIONS = frozenset({
+    "continue",
+    "handle this",
+    "fix it",
+    "follow up",
+    "follow-up",
+    "see above",
+    "see review",
+    "tbd",
+    "todo",
+    "as needed",
+    "proceed",
+    "next steps",
+    "will check",
+    "investigate",
+})
+
+_FULL_SHA_RE = re.compile(r"\b[0-9a-f]{40}\b", re.IGNORECASE)
+_SHORT_SHA_RE = re.compile(r"\b[0-9a-f]{7,40}\b", re.IGNORECASE)
+_PR_REF_RE = re.compile(r"(?:PR\s*#|pull\s*#)\d+|\b#\d{2,}\b", re.IGNORECASE)
+_APPROVAL_PROOF_RE = re.compile(
+    r"\b(?:approved|approval_at_current_head|APPROVE)\b",
+    re.IGNORECASE,
+)
+_UNBLOCK_RE = re.compile(
+    r"\b(?:unblock|until|after|once|when|requires?|must)\b",
+    re.IGNORECASE,
+)
+
+
+def _parse_fields(body: str) -> dict[str, str]:
+    """Parse KEY: value fields, including values continued on following lines."""
+    fields: dict[str, str] = {}
+    current_key: str | None = None
+    current_lines: list[str] = []
+
+    def _flush() -> None:
+        nonlocal current_key, current_lines
+        if current_key is not None:
+            fields[current_key] = "\n".join(current_lines).strip()
+        current_key = None
+        current_lines = []
+
+    for line in (body or "").splitlines():
+        if line.startswith("## "):
+            _flush()
+            continue
+        match = re.match(r"^([A-Z][A-Z0-9_]*)\s*:\s*(.*)$", line)
+        if match:
+            _flush()
+            current_key = match.group(1).strip().upper()
+            rest = match.group(2)
+            current_lines = [rest] if rest else []
+        elif current_key is not None:
+            current_lines.append(line)
+    _flush()
+    return fields
+
+
+def _is_machine_generated(body: str) -> bool:
+    text = body or ""
+    return any(marker in text for marker in _MACHINE_MARKERS)
+
+
+def _has_canonical_heading(body: str) -> bool:
+    return any(h in (body or "") for h in _CANONICAL_HEADINGS)
+
+
+def is_workflow_changing_comment(body: str) -> bool:
+    """True when comment text implies a workflow/state transition."""
+    text = (body or "").strip()
+    if not text:
+        return False
+    if _is_machine_generated(text):
+        return False
+    if _has_canonical_heading(text):
+        return True
+    if _WORKFLOW_TRIGGERS.search(text):
+        return True
+    fields = _parse_fields(text)
+    if "STATE" in fields or "WHO_IS_NEXT" in fields:
+        return True
+    return False
+
+
+def infer_comment_context(body: str, *, explicit: str | None = None) -> str:
+    if explicit:
+        return explicit
+    text = body or ""
+    if "## Canonical Discussion Summary" in text:
+        return "discussion_summary"
+    if "## Canonical PR State" in text:
+        return "pr_comment"
+    if "## Canonical Issue State" in text:
+        return "issue_comment"
+    fields = _parse_fields(text)
+    if fields.get("CANONICAL_ITEM") or fields.get("SUPERSEDED_ITEM"):
+        return "supersession"
+    if any(k in fields for k in ("HEAD_SHA", "REVIEW_STATUS", "MERGE_READY", "ISSUE")):
+        return "pr_comment"
+    return "issue_comment"
+
+
+def _required_fields_for_context(context: str) -> tuple[str, ...]:
+    if context in ("pr_comment", "pr_review"):
+        return _PR_REQUIRED
+    if context == "supersession":
+        return _SUPERSESSION_REQUIRED
+    if context == "discussion_summary":
+        return _BASE_REQUIRED + ("DECISION", "SUBSTANTIVE_COMMENTS")
+    return _ISSUE_REQUIRED
+
+
+def _is_vague_next_action(value: str) -> bool:
+    normalized = re.sub(r"\s+", " ", (value or "").strip().lower())
+    normalized = normalized.rstrip(".")
+    if not normalized:
+        return True
+    if normalized in _VAGUE_NEXT_ACTIONS:
+        return True
+    if len(normalized) < 12:
+        return True
+    return False
+
+
+def _next_prompt_ok(value: str) -> bool:
+    text = (value or "").strip()
+    if len(text) < 40:
+        return False
+    if text.lower() in {"n/a", "none", "tbd", "todo"}:
+        return False
+    return True
+
+
+def _state_value(fields: dict[str, str]) -> str:
+    return (fields.get("STATE") or "").strip().lower()
+
+
+def _suggested_template(context: str) -> str:
+    if context == "pr_comment" or context == "pr_review":
+        return (
+            "## Canonical PR State\n\n"
+            "STATE:\n"
+            "WHO_IS_NEXT:\n"
+            "NEXT_ACTION:\n"
+            "NEXT_PROMPT:\n"
+            "```text\n<paste-ready prompt>\n```\n"
+            "WHAT_HAPPENED:\n"
+            "WHY:\n"
+            "ISSUE:\n"
+            "HEAD_SHA:\n"
+            "REVIEW_STATUS:\n"
+            "MERGE_READY:\n"
+            "BLOCKERS:\n"
+            "VALIDATION:\n"
+            "LAST_UPDATED_BY:\n"
+        )
+    if context == "supersession":
+        return (
+            "## Canonical PR State\n\n"
+            "STATE:\nsuperseded\n"
+            "WHO_IS_NEXT:\nreconciler\n"
+            "NEXT_ACTION:\n"
+            "NEXT_PROMPT:\n"
+            "WHY:\n"
+            "CANONICAL_ITEM:\n"
+            "SUPERSEDED_ITEM:\n"
+            "CLOSE_OR_KEEP_OPEN:\n"
+        )
+    if context == "discussion_summary":
+        return (
+            "## Canonical Discussion Summary\n\n"
+            "STATE:\n"
+            "WHO_IS_NEXT:\n"
+            "DECISION:\n"
+            "WHY:\n"
+            "SUBSTANTIVE_COMMENTS:\n"
+            "NEXT_ACTION:\n"
+            "NEXT_PROMPT:\n"
+        )
+    return (
+        "## Canonical Issue State\n\n"
+        "STATE:\n"
+        "WHO_IS_NEXT:\n"
+        "NEXT_ACTION:\n"
+        "NEXT_PROMPT:\n"
+        "```text\n<paste-ready prompt>\n```\n"
+        "WHAT_HAPPENED:\n"
+        "WHY:\n"
+        "RELATED_PRS:\n"
+        "BLOCKERS:\n"
+        "VALIDATION:\n"
+        "LAST_UPDATED_BY:\n"
+    )
+
+
+def _build_correction_message(
+    *,
+    missing_fields: list[str],
+    vague_fields: list[str],
+    extra_reasons: list[str],
+    context: str,
+) -> str:
+    parts = [
+        "Canonical comment validation failed (fail closed before posting).",
+    ]
+    if missing_fields:
+        parts.append("Missing fields: " + ", ".join(missing_fields) + ".")
+    if vague_fields:
+        parts.append("Vague or invalid fields: " + ", ".join(vague_fields) + ".")
+    parts.extend(extra_reasons)
+    parts.append("Fill the suggested template and retry.")
+    parts.append("Suggested template:\n" + _suggested_template(context))
+    return "\n".join(parts)
+
+
+def assess_canonical_comment(
+    body: str,
+    *,
+    context: str | None = None,
+    force_workflow: bool = False,
+) -> dict[str, Any]:
+    """Validate outgoing comment text before a Gitea mutation."""
+    text = (body or "").strip()
+    ctx = infer_comment_context(text, explicit=context)
+
+    if not text:
+        return {
+            "allowed": True,
+            "is_workflow_comment": False,
+            "context": ctx,
+            "missing_fields": [],
+            "vague_fields": [],
+            "correction_message": "",
+            "suggested_template": "",
+        }
+
+    if _is_machine_generated(text):
+        return {
+            "allowed": True,
+            "is_workflow_comment": False,
+            "context": ctx,
+            "missing_fields": [],
+            "vague_fields": [],
+            "correction_message": "",
+            "suggested_template": "",
+        }
+
+    workflow = force_workflow or is_workflow_changing_comment(text)
+    if not workflow:
+        return {
+            "allowed": True,
+            "is_workflow_comment": False,
+            "context": ctx,
+            "missing_fields": [],
+            "vague_fields": [],
+            "correction_message": "",
+            "suggested_template": "",
+        }
+
+    fields = _parse_fields(text)
+    required = _required_fields_for_context(ctx)
+    missing = [name for name in required if not (fields.get(name) or "").strip()]
+
+    vague: list[str] = []
+    extra: list[str] = []
+
+    who = (fields.get("WHO_IS_NEXT") or "").strip().lower()
+    if who and who not in VALID_ROLES:
+        vague.append("WHO_IS_NEXT")
+        extra.append(
+            f"WHO_IS_NEXT must be one of: {', '.join(sorted(VALID_ROLES))}."
+        )
+
+    next_action = fields.get("NEXT_ACTION") or ""
+    if next_action and _is_vague_next_action(next_action):
+        vague.append("NEXT_ACTION")
+
+    next_prompt = fields.get("NEXT_PROMPT") or ""
+    if not _next_prompt_ok(next_prompt):
+        if "NEXT_PROMPT" not in missing:
+            vague.append("NEXT_PROMPT")
+
+    state = _state_value(fields)
+    blockers = (fields.get("BLOCKERS") or "").strip()
+    if "blocked" in state:
+        if not blockers or blockers.lower() in {"none", "n/a"}:
+            missing.append("BLOCKERS (unblock condition)")
+        elif not _UNBLOCK_RE.search(blockers):
+            vague.append("BLOCKERS")
+            extra.append("BLOCKED state requires an explicit unblock condition in BLOCKERS.")
+
+    if "superseded" in state:
+        canon = (fields.get("CANONICAL_ITEM") or fields.get("SUPERSEDED_BY") or "").strip()
+        superseded = (fields.get("SUPERSEDED_ITEM") or fields.get("SUPERSEDES") or "").strip()
+        if not canon and "CANONICAL_ITEM" not in missing:
+            missing.append("CANONICAL_ITEM")
+        if not superseded and "SUPERSEDED_ITEM" not in missing:
+            missing.append("SUPERSEDED_ITEM")
+
+    if "ready-to-merge" in state or "ready to merge" in state:
+        head_sha = fields.get("HEAD_SHA") or ""
+        merge_ready = fields.get("MERGE_READY") or ""
+        review_status = fields.get("REVIEW_STATUS") or ""
+        validation = fields.get("VALIDATION") or ""
+        proof_blob = " ".join((head_sha, merge_ready, review_status, validation))
+        has_sha = bool(_FULL_SHA_RE.search(proof_blob) or _SHORT_SHA_RE.search(proof_blob))
+        has_approval = bool(_APPROVAL_PROOF_RE.search(proof_blob))
+        if not has_sha or not has_approval:
+            extra.append(
+                "ready-to-merge STATE requires approval proof and HEAD_SHA in "
+                "REVIEW_STATUS, MERGE_READY, HEAD_SHA, or VALIDATION."
+            )
+
+    if ctx in ("pr_comment", "pr_review"):
+        head_sha = (fields.get("HEAD_SHA") or "").strip()
+        if not head_sha or not _SHORT_SHA_RE.search(head_sha):
+            if "HEAD_SHA" not in missing:
+                missing.append("HEAD_SHA")
+
+    if ctx == "issue_comment" and _PR_REF_RE.search(text):
+        related = (fields.get("RELATED_PRS") or "").strip()
+        if not related or related.lower() in {"none", "n/a", "-"}:
+            missing.append("RELATED_PRS")
+
+    # #695 AC7: untrusted / offline approval claims cannot certify merger handoff.
+    try:
+        import review_quarantine as _rq
+
+        for claim_reason in _rq.assess_untrusted_canonical_approval_claim(text):
+            extra.append(claim_reason)
+    except Exception:
+        # Fail closed on import/runtime errors for approval-shaped claims only.
+        lower = text.lower()
+        if (
+            "state:\napproved" in lower
+            or "who_is_next:\nmerger" in lower
+            or "merge_ready: true" in lower
+            or "merge_ready:\ntrue" in lower
+            or "ready-to-merge" in lower
+        ):
+            extra.append(
+                "canonical approval/merge-ready claim could not be verified "
+                "for native review proof (#695 AC7; fail closed)"
+            )
+
+    allowed = not missing and not vague and not extra
+    correction = ""
+    if not allowed:
+        correction = _build_correction_message(
+            missing_fields=missing,
+            vague_fields=vague,
+            extra_reasons=extra,
+            context=ctx,
+        )
+
+    return {
+        "allowed": allowed,
+        "is_workflow_comment": True,
+        "context": ctx,
+        "missing_fields": missing,
+        "vague_fields": vague,
+        "extra_reasons": extra,
+        "correction_message": correction,
+        "suggested_template": _suggested_template(ctx) if not allowed else "",
+    }

canonical_state_comments.py

@@ -0,0 +1,171 @@
+"""Canonical state comment validation helpers (#495).
+
+These helpers validate durable issue/PR/discussion state handoff comments.
+They are intentionally pure and do not post comments; MCP mutation hooks are
+owned by #496.
+"""
+
+from __future__ import annotations
+
+import re
+
+CANONICAL_HEADINGS = (
+    "canonical issue state",
+    "canonical pr state",
+    "canonical discussion summary",
+)
+
+REQUIRED_FIELDS = ("STATE", "WHO_IS_NEXT", "NEXT_ACTION", "NEXT_PROMPT")
+ALLOWED_NEXT_ACTORS = {
+    "controller",
+    "author",
+    "reviewer",
+    "merger",
+    "reconciler",
+    "user",
+}
+
+VAGUE_NEXT_ACTIONS = {
+    "continue",
+    "handle this",
+    "do it",
+    "fix it",
+    "proceed",
+    "follow up",
+    "next",
+    "tbd",
+    "todo",
+    "n/a",
+    "none",
+}
+
+_FIELD_RE = re.compile(r"^\s*(?:[-*]\s*)?([A-Z][A-Z0-9_ ]+)\s*:\s*(.*)$")
+_FULL_SHA_RE = re.compile(r"\b[0-9a-f]{40}\b", re.IGNORECASE)
+_CLAIMS_STATE_UPDATE_RE = re.compile(
+    r"canonical\s+(?:issue|pr|discussion)?\s*state|"
+    r"state\s+comment\s+(?:posted|created|updated)|"
+    r"next[- ]action\s+comment\s+(?:posted|created|updated)",
+    re.IGNORECASE,
+)
+
+
+def extract_state_fields(text: str | None) -> dict[str, str]:
+    """Return upper-case labeled fields from a canonical state block."""
+    fields: dict[str, str] = {}
+    current_key: str | None = None
+    for line in (text or "").splitlines():
+        match = _FIELD_RE.match(line)
+        if match:
+            current_key = match.group(1).strip().upper().replace(" ", "_")
+            fields[current_key] = match.group(2).strip()
+            continue
+        stripped = line.strip()
+        if current_key and stripped and not stripped.startswith("#"):
+            existing = fields.get(current_key, "")
+            fields[current_key] = (
+                f"{existing}\n{stripped}" if existing else stripped
+            )
+    return fields
+
+
+def contains_canonical_state_block(text: str | None) -> bool:
+    lower = (text or "").lower()
+    return any(heading in lower for heading in CANONICAL_HEADINGS)
+
+
+def claims_canonical_state_update(text: str | None) -> bool:
+    """Return True when text claims a canonical state/next-action update."""
+    return bool(_CLAIMS_STATE_UPDATE_RE.search(text or ""))
+
+
+def _empty_or_placeholder(value: str | None) -> bool:
+    value = (value or "").strip().lower()
+    return not value or value in {"none", "n/a", "unknown", "tbd", "<...>"}
+
+
+def _vague_next_action(value: str | None) -> bool:
+    normalized = re.sub(r"\s+", " ", (value or "").strip().lower())
+    return normalized in VAGUE_NEXT_ACTIONS
+
+
+def validate_canonical_state_comment(text: str | None) -> dict:
+    """Validate a canonical state comment or embedded final-report block.
+
+    Returns a dict with ``valid`` and ``reasons``. The validator focuses on
+    fields that make continuation possible: current state, next actor, next
+    action, and paste-ready next prompt, plus a few contradiction checks.
+    """
+    fields = extract_state_fields(text)
+    reasons: list[str] = []
+
+    for field in REQUIRED_FIELDS:
+        if _empty_or_placeholder(fields.get(field)):
+            reasons.append(f"missing required canonical state field: {field}")
+
+    actor = (fields.get("WHO_IS_NEXT") or "").strip().lower()
+    if actor and actor not in ALLOWED_NEXT_ACTORS:
+        reasons.append(
+            "WHO_IS_NEXT must be one of: "
+            + ", ".join(sorted(ALLOWED_NEXT_ACTORS))
+        )
+
+    if _vague_next_action(fields.get("NEXT_ACTION")):
+        reasons.append("NEXT_ACTION is too vague for durable continuation")
+
+    state = (fields.get("STATE") or "").strip().lower().replace("-", "_")
+    if "ready_to_merge" in state or (
+        state == "approved" and "pr" in (text or "").lower()
+    ):
+        review_status = (fields.get("REVIEW_STATUS") or "").lower()
+        head_sha = fields.get("HEAD_SHA") or ""
+        merge_ready = (fields.get("MERGE_READY") or "").lower()
+        if "approved" not in review_status:
+            reasons.append("ready-to-merge state requires approved REVIEW_STATUS")
+        if not _FULL_SHA_RE.search(head_sha):
+            reasons.append("ready-to-merge state requires full HEAD_SHA proof")
+        if merge_ready and not merge_ready.startswith(("yes", "true")):
+            reasons.append("ready-to-merge state contradicts MERGE_READY")
+
+    if "superseded" in state:
+        canonical = fields.get("CANONICAL_ITEM") or fields.get("SUPERSEDED_BY")
+        if _empty_or_placeholder(canonical):
+            reasons.append("superseded state requires canonical item proof")
+
+    if "blocked" in state:
+        blockers = fields.get("BLOCKERS") or ""
+        if _empty_or_placeholder(blockers):
+            reasons.append("blocked state requires BLOCKERS/unblock condition")
+
+    return {
+        "valid": not reasons,
+        "fields": fields,
+        "reasons": reasons,
+    }
+
+
+def validate_final_report_state_update(report_text: str | None) -> dict:
+    """Validate canonical state-update claims inside a final report."""
+    text = report_text or ""
+    if not claims_canonical_state_update(text) and not contains_canonical_state_block(text):
+        return {
+            "applicable": False,
+            "valid": True,
+            "reasons": [],
+        }
+
+    if not contains_canonical_state_block(text):
+        return {
+            "applicable": True,
+            "valid": False,
+            "reasons": [
+                "final report claims a canonical state update but includes no canonical state block"
+            ],
+        }
+
+    result = validate_canonical_state_comment(text)
+    return {
+        "applicable": True,
+        "valid": result["valid"],
+        "fields": result["fields"],
+        "reasons": result["reasons"],
+    }

canonical_thread_handoff.py

@@ -0,0 +1,213 @@
+"""Canonical Thread Handoff (CTH) protocol for issue and PR comments (#505).
+
+A CTH comment is the authoritative workflow handoff in a Gitea issue or PR
+thread. It records current state, decisions, blockers, proof, and the exact
+next prompt/action for the next LLM or person.
+"""
+
+from __future__ import annotations
+
+import re
+from datetime import datetime, timezone
+from typing import Any
+
+MARKER = "<!-- cth:v1 -->"
+
+CTH_TERM = "Canonical Thread Handoff"
+CTH_ABBREV = "CTH"
+
+CTH_TYPES = frozenset({
+    "State Handoff",
+    "Controller Decision",
+    "Author Handoff",
+    "Reviewer Handoff",
+    "Merger Handoff",
+    "Supersession Notice",
+    "Blocker",
+})
+
+_REQUIRED_BASE_FIELDS = (
+    "status",
+    "next owner",
+    "current blocker",
+    "decision",
+    "proof",
+    "next action",
+    "ready-to-paste prompt",
+)
+
+_HEADING_RE = re.compile(
+    r"^##\s+CTH:\s*(.+?)\s*$",
+    re.IGNORECASE | re.MULTILINE,
+)
+_FIELD_RE = re.compile(
+    r"^([A-Za-z][A-Za-z0-9 /-]*):\s*(.+?)\s*$",
+    re.MULTILINE,
+)
+
+
+def format_cth_body(
+    *,
+    cth_type: str,
+    status: str,
+    next_owner: str,
+    current_blocker: str = "none",
+    decision: str = "none",
+    proof: str = "none",
+    next_action: str = "none",
+    ready_to_paste_prompt: str = "none",
+    extra_fields: dict[str, str] | None = None,
+) -> str:
+    """Render a canonical CTH comment body."""
+    normalized_type = (cth_type or "").strip()
+    if normalized_type not in CTH_TYPES:
+        raise ValueError(
+            f"unknown CTH type '{cth_type}'; expected one of {sorted(CTH_TYPES)}"
+        )
+    lines = [
+        MARKER,
+        f"## CTH: {normalized_type}",
+        "",
+        f"Status: {(status or 'unknown').strip() or 'unknown'}",
+        f"Next owner: {(next_owner or 'unknown').strip() or 'unknown'}",
+        f"Current blocker: {(current_blocker or 'none').strip() or 'none'}",
+        f"Decision: {(decision or 'none').strip() or 'none'}",
+        f"Proof: {(proof or 'none').strip() or 'none'}",
+        f"Next action: {(next_action or 'none').strip() or 'none'}",
+        f"Ready-to-paste prompt: {(ready_to_paste_prompt or 'none').strip() or 'none'}",
+    ]
+    for key, value in (extra_fields or {}).items():
+        label = (key or "").strip()
+        if not label:
+            continue
+        lines.append(f"{label}: {(value or 'none').strip() or 'none'}")
+    return "\n".join(lines)
+
+
+def parse_cth_comment(body: str) -> dict[str, Any] | None:
+    """Parse one CTH comment body, or None when not a CTH comment."""
+    text = body or ""
+    if MARKER not in text and not _HEADING_RE.search(text):
+        return None
+    heading = _HEADING_RE.search(text)
+    if not heading:
+        return None
+    cth_type = heading.group(1).strip()
+    fields: dict[str, str] = {}
+    for match in _FIELD_RE.finditer(text):
+        key = match.group(1).strip().lower()
+        if key.startswith("cth"):
+            continue
+        fields[key] = match.group(2).strip()
+    return {
+        "cth_type": cth_type,
+        "fields": fields,
+        "raw_body": text,
+    }
+
+
+def assess_cth_comment(body: str) -> dict[str, Any]:
+    """Validate a CTH comment has required base fields and a known type."""
+    parsed = parse_cth_comment(body)
+    reasons: list[str] = []
+    if not parsed:
+        return {
+            "valid": False,
+            "block": True,
+            "reasons": ["comment is not a Canonical Thread Handoff (CTH)"],
+            "parsed": None,
+        }
+
+    cth_type = parsed.get("cth_type") or ""
+    if cth_type not in CTH_TYPES:
+        reasons.append(
+            f"unknown CTH type '{cth_type}'; expected one of {sorted(CTH_TYPES)}"
+        )
+
+    fields = parsed.get("fields") or {}
+    missing = [name for name in _REQUIRED_BASE_FIELDS if not (fields.get(name) or "").strip()]
+    if missing:
+        reasons.append(
+            "CTH missing required fields: " + ", ".join(missing)
+        )
+
+    vague_prompt = (fields.get("ready-to-paste prompt") or "").strip().lower()
+    if vague_prompt in {"", "none", "tbd", "n/a", "todo"}:
+        reasons.append("ready-to-paste prompt must be concrete and paste-ready")
+
+    return {
+        "valid": not reasons,
+        "block": bool(reasons),
+        "reasons": reasons,
+        "parsed": parsed,
+        "cth_type": cth_type,
+    }
+
+
+def find_latest_cth(comments: list[dict[str, Any]]) -> dict[str, Any] | None:
+    """Return the newest valid CTH comment from a Gitea comment list."""
+    latest: dict[str, Any] | None = None
+    latest_ts: datetime | None = None
+    for comment in comments or []:
+        body = comment.get("body") or ""
+        parsed = parse_cth_comment(body)
+        if not parsed:
+            continue
+        created = _parse_timestamp(comment.get("created_at"))
+        if latest is None or (created and (latest_ts is None or created >= latest_ts)):
+            latest = {
+                "comment_id": comment.get("id"),
+                "created_at": comment.get("created_at"),
+                "author": (comment.get("user") or {}).get("login"),
+                **parsed,
+            }
+            latest_ts = created
+    return latest
+
+
+def assess_cth_supersedes_non_cth(
+    *,
+    latest_cth: dict[str, Any] | None,
+    narrative_claim: str,
+) -> dict[str, Any]:
+    """Fail closed when narrative relies on stale non-CTH state despite newer CTH."""
+    if not latest_cth:
+        return {"block": False, "reasons": []}
+    text = (narrative_claim or "").strip()
+    if not text:
+        return {"block": False, "reasons": []}
+    if parse_cth_comment(text):
+        return {"block": False, "reasons": []}
+    return {
+        "block": True,
+        "reasons": [
+            "workflow narrative must treat the latest CTH comment as authoritative; "
+            f"found newer CTH type '{latest_cth.get('cth_type')}' "
+            f"(comment #{latest_cth.get('comment_id')})"
+        ],
+    }
+
+
+def _parse_timestamp(value: str | None) -> datetime | None:
+    if not value:
+        return None
+    text = value.strip()
+    if text.endswith("Z"):
+        text = text[:-1] + "+00:00"
+    try:
+        parsed = datetime.fromisoformat(text)
+    except ValueError:
+        return None
+    if parsed.tzinfo is None:
+        return parsed.replace(tzinfo=timezone.utc)
+    return parsed.astimezone(timezone.utc)
+
+
+EXAMPLE_SCENARIOS = (
+    "pr_approved_ready_for_merger",
+    "pr_request_changes_to_author",
+    "duplicate_superseded_pr_closure",
+    "blocked_dirty_root_worktree",
+    "stale_head_requires_fresh_review",
+    "issue_implementation_handoff",
+)

conflict_fix_classification.py

@@ -0,0 +1,266 @@
+"""Live PR head re-pin before conflict-fix classification (#522).
+
+Open PR inventory fields (mergeable, head_sha) can be stale. Author sessions
+must re-fetch live PR state and pin the live head before classifying a PR as
+conflicted or creating a conflict-fix worktree.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+_FULL_SHA = re.compile(r"^[0-9a-f]{40}$", re.IGNORECASE)
+
+_INVENTORY_HEAD_RE = re.compile(
+    r"(?:inventory head sha|stale inventory head sha)\s*:\s*([0-9a-f]{40})",
+    re.IGNORECASE,
+)
+_LIVE_HEAD_RE = re.compile(
+    r"(?:live head sha|pinned head sha|live pr head sha)\s*:\s*([0-9a-f]{40})",
+    re.IGNORECASE,
+)
+_REPINS_TOOL_RE = re.compile(
+    r"gitea_assess_conflict_fix_classification",
+    re.IGNORECASE,
+)
+_CLASSIFICATION_RE = re.compile(
+    r"conflict[- ]fix classification\s*:\s*(\S+)",
+    re.IGNORECASE,
+)
+
+CLASSIFICATION_STALE_INVENTORY_SKIP = "stale_inventory_skip"
+CLASSIFICATION_CONFLICT_FIX_NEEDED = "conflict_fix_needed"
+CLASSIFICATION_LIVE_MERGEABLE = "live_mergeable_skip"
+CLASSIFICATION_INCOMPLETE = "incomplete_live_repin"
+
+_VALID_CLASSIFICATIONS = frozenset({
+    CLASSIFICATION_STALE_INVENTORY_SKIP,
+    CLASSIFICATION_CONFLICT_FIX_NEEDED,
+    CLASSIFICATION_LIVE_MERGEABLE,
+    CLASSIFICATION_INCOMPLETE,
+})
+
+
+def _normalize_sha(value: str | None) -> str | None:
+    text = (value or "").strip().lower()
+    if not text:
+        return None
+    return text if _FULL_SHA.match(text) else None
+
+
+def assess_conflict_fix_classification(
+    *,
+    pr_number: int,
+    inventory_head_sha: str | None = None,
+    inventory_mergeable: bool | None = None,
+    live_head_sha: str | None = None,
+    live_mergeable: bool | None = None,
+) -> dict[str, Any]:
+    """Classify whether conflict-fix work is justified from live PR state.
+
+    Inventory fields are advisory only. Live head + live mergeable are required
+    before any conflict-fix worktree may be created.
+    """
+    inv_head = _normalize_sha(inventory_head_sha)
+    live_head = _normalize_sha(live_head_sha)
+    reasons: list[str] = []
+
+    if not isinstance(pr_number, int) or pr_number <= 0:
+        return {
+            "classification": CLASSIFICATION_INCOMPLETE,
+            "worktree_allowed": False,
+            "skip_author_mutation": True,
+            "inventory_stale_head": False,
+            "inventory_stale_mergeable": False,
+            "pinned_head_sha": None,
+            "inventory_head_sha": inv_head,
+            "live_head_sha": live_head,
+            "inventory_mergeable": inventory_mergeable,
+            "live_mergeable": live_mergeable,
+            "pr_number": pr_number,
+            "reasons": ["pr_number must be a positive integer (fail closed)"],
+        }
+
+    if live_head is None:
+        reasons.append(
+            "live PR head SHA missing or not a full 40-char hex SHA; "
+            "re-fetch the PR before conflict-fix classification (fail closed)"
+        )
+    if live_mergeable is None:
+        reasons.append(
+            "live PR mergeable value missing; re-fetch the PR before "
+            "conflict-fix classification (fail closed)"
+        )
+
+    if reasons:
+        return {
+            "classification": CLASSIFICATION_INCOMPLETE,
+            "worktree_allowed": False,
+            "skip_author_mutation": True,
+            "inventory_stale_head": bool(
+                inv_head and live_head and inv_head != live_head
+            ),
+            "inventory_stale_mergeable": (
+                inventory_mergeable is not None
+                and live_mergeable is not None
+                and bool(inventory_mergeable) != bool(live_mergeable)
+            ),
+            "pinned_head_sha": live_head,
+            "inventory_head_sha": inv_head,
+            "live_head_sha": live_head,
+            "inventory_mergeable": inventory_mergeable,
+            "live_mergeable": live_mergeable,
+            "pr_number": pr_number,
+            "reasons": reasons,
+        }
+
+    inventory_stale_head = bool(inv_head and inv_head != live_head)
+    inventory_stale_mergeable = (
+        inventory_mergeable is not None
+        and bool(inventory_mergeable) != bool(live_mergeable)
+    )
+
+    # Live mergeable wins: do not start conflict-fix work.
+    if live_mergeable is True:
+        classification = (
+            CLASSIFICATION_STALE_INVENTORY_SKIP
+            if (
+                inventory_mergeable is False
+                or inventory_stale_head
+                or inventory_stale_mergeable
+            )
+            else CLASSIFICATION_LIVE_MERGEABLE
+        )
+        note = []
+        if inventory_stale_head:
+            note.append(
+                f"inventory head {inv_head} differs from live head {live_head}; "
+                "use live head only"
+            )
+        if inventory_mergeable is False:
+            note.append(
+                "inventory reported mergeable:false but live PR is mergeable:true; "
+                "skip author conflict-fix mutation"
+            )
+        return {
+            "classification": classification,
+            "worktree_allowed": False,
+            "skip_author_mutation": True,
+            "inventory_stale_head": inventory_stale_head,
+            "inventory_stale_mergeable": inventory_stale_mergeable,
+            "pinned_head_sha": live_head,
+            "inventory_head_sha": inv_head,
+            "live_head_sha": live_head,
+            "inventory_mergeable": inventory_mergeable,
+            "live_mergeable": live_mergeable,
+            "pr_number": pr_number,
+            "reasons": note,
+        }
+
+    # live_mergeable is False → conflict-fix may proceed on the pinned live head.
+    note = []
+    if inventory_stale_head:
+        note.append(
+            f"inventory head {inv_head} differs from live head {live_head}; "
+            "pin and use live head only for conflict-fix work"
+        )
+    if inventory_mergeable is True:
+        note.append(
+            "inventory reported mergeable:true but live PR is mergeable:false; "
+            "trust live mergeable and proceed only with live head pin"
+        )
+    note.append(
+        f"live PR #{pr_number} is mergeable:false at pinned head {live_head}; "
+        "conflict-fix worktree allowed for that head only"
+    )
+    return {
+        "classification": CLASSIFICATION_CONFLICT_FIX_NEEDED,
+        "worktree_allowed": True,
+        "skip_author_mutation": False,
+        "inventory_stale_head": inventory_stale_head,
+        "inventory_stale_mergeable": inventory_stale_mergeable,
+        "pinned_head_sha": live_head,
+        "inventory_head_sha": inv_head,
+        "live_head_sha": live_head,
+        "inventory_mergeable": inventory_mergeable,
+        "live_mergeable": live_mergeable,
+        "pr_number": pr_number,
+        "reasons": note,
+    }
+
+
+def assess_conflict_fix_classification_final_report(
+    report_text: str,
+    **_kwargs: Any,
+) -> dict[str, Any]:
+    """Require live-head re-pin proof when a report claims conflict-fix work (#522)."""
+    text = report_text or ""
+    lower = text.lower()
+    mentions_conflict_fix = (
+        "conflict-fix" in lower
+        or "conflict fix" in lower
+        or "conflict_fix" in lower
+    )
+    if not mentions_conflict_fix:
+        return {"proven": True, "reasons": [], "applicable": False}
+
+    reasons: list[str] = []
+    if not _REPINS_TOOL_RE.search(text):
+        reasons.append(
+            "conflict-fix report must cite gitea_assess_conflict_fix_classification "
+            "(live head re-pin tool)"
+        )
+
+    live_match = _LIVE_HEAD_RE.search(text)
+    if not live_match:
+        reasons.append(
+            "conflict-fix report must state Live head SHA: <40-char hex> "
+            "(or Pinned head SHA / Live PR head SHA)"
+        )
+    else:
+        live_sha = _normalize_sha(live_match.group(1))
+        if live_sha is None:
+            reasons.append("live head SHA is not a full 40-char hex digest")
+
+    inv_match = _INVENTORY_HEAD_RE.search(text)
+    # Inventory head is optional but recommended when classification is stale skip.
+    class_match = _CLASSIFICATION_RE.search(text)
+    if not class_match:
+        reasons.append(
+            "conflict-fix report must state Conflict-fix classification: "
+            f"<{'|'.join(sorted(_VALID_CLASSIFICATIONS))}>"
+        )
+    else:
+        classification = class_match.group(1).strip().lower().replace(" ", "_")
+        # allow hyphenated forms
+        classification = classification.replace("-", "_")
+        if classification not in _VALID_CLASSIFICATIONS:
+            reasons.append(
+                f"unknown conflict-fix classification {class_match.group(1)!r}; "
+                f"expected one of {sorted(_VALID_CLASSIFICATIONS)}"
+            )
+
+    if inv_match and live_match:
+        inv_sha = _normalize_sha(inv_match.group(1))
+        live_sha = _normalize_sha(live_match.group(1))
+        if inv_sha and live_sha and inv_sha != live_sha:
+            # Require explicit note that live head was used.
+            if "use live head" not in lower and "live head only" not in lower:
+                reasons.append(
+                    "inventory head differs from live head; report must state that "
+                    "the live head was used exclusively"
+                )
+
+    return {
+        "proven": not reasons,
+        "reasons": reasons,
+        "applicable": True,
+        "inventory_head_sha": _normalize_sha(inv_match.group(1)) if inv_match else None,
+        "live_head_sha": _normalize_sha(live_match.group(1)) if live_match else None,
+        "classification": (
+            class_match.group(1).strip().lower().replace("-", "_").replace(" ", "_")
+            if class_match
+            else None
+        ),
+    }

controller_closure_baseline_proof.py

@@ -0,0 +1,63 @@
+"""Controller-closure baseline-proof gate (#529, criterion 6).
+
+A controller (or any session) may close a tracking issue with a closure
+report that summarizes validation. When that report calls a non-zero
+test-suite exit an "expected pre-existing failure" (or baseline / known
+failure) it must carry pre-merge proof; otherwise the closure buries an
+unproven regression as an accepted baseline and weakens controller
+confidence in the durable state.
+
+This module fails closed on exactly that case by reusing the pre-merge
+baseline proof verifier (#533). A closure report is allowed when it is
+empty (no validation claim), a clean pass, or a baseline failure proven on
+the PR pre-merge base commit (or a documented known-failure record that
+predates the PR).
+"""
+
+from __future__ import annotations
+
+from typing import Any
+
+from premerge_baseline_proof import CLEAN_PASS, assess_premerge_baseline_proof
+
+
+def assess_controller_closure_baseline_proof(
+    closure_report: str | None,
+) -> dict[str, Any]:
+    """Assess whether an issue-closure report may proceed.
+
+    Returns a dict with ``block``, ``proven``, ``label``, ``reasons``,
+    ``skipped`` and ``safe_next_action``. Only blocks when the closure
+    report claims a non-zero validation exit is an expected
+    pre-existing/baseline failure without valid pre-merge proof.
+    """
+    text = (closure_report or "").strip()
+    if not text:
+        return {
+            "block": False,
+            "proven": True,
+            "label": CLEAN_PASS,
+            "reasons": [],
+            "skipped": True,
+            "safe_next_action": "",
+        }
+
+    result = assess_premerge_baseline_proof(text)
+    block = bool(result.get("block"))
+    return {
+        "block": block,
+        "proven": not block,
+        "label": result.get("label"),
+        "reasons": list(result.get("reasons") or []),
+        "skipped": bool(result.get("skipped", False)),
+        "safe_next_action": (
+            result.get("safe_next_action")
+            or (
+                "provide pre-merge baseline proof (base commit, tested commit, "
+                "command, exit status, failure signature) before closing on an "
+                "expected pre-existing failure"
+            )
+        )
+        if block
+        else "",
+    }

create_issue.py

@@ -29,6 +29,7 @@ from gitea_auth import (
     get_credentials, resolve_remote, add_remote_args,
     api_request, repo_api_url,
 )
+import issue_workflow_labels
 
 
 def main(argv=None):
@@ -38,6 +39,16 @@ def main(argv=None):
     parser.add_argument("--body", default="", help="Issue body text.")
     parser.add_argument("--body-file",
                         help="Read issue body from this file ('-' for stdin).")
+    parser.add_argument("--label", action="append", default=[],
+                        help="Existing label name to apply; may be repeated.")
+    parser.add_argument("--type-label",
+                        help="Issue type, e.g. feature or type:feature.")
+    parser.add_argument("--status-label",
+                        help="Initial status, e.g. ready or status:ready.")
+    parser.add_argument("--discussion", action="store_true",
+                        help="Apply type:discussion.")
+    parser.add_argument("--require-workflow-labels", action="store_true",
+                        help="Fail unless one type:* and one status:* label are supplied.")
     args = parser.parse_args(argv)
 
     host, org, repo = resolve_remote(args)
@@ -50,6 +61,24 @@ def main(argv=None):
             with open(args.body_file, "r", encoding="utf-8") as fh:
                 body = fh.read()
 
+    requested_labels = issue_workflow_labels.labels_for_new_issue(
+        issue_type=args.type_label,
+        initial_status=args.status_label,
+        extra_labels=args.label,
+        discussion=args.discussion,
+    )
+    label_assessment = issue_workflow_labels.assess_issue_labels(
+        requested_labels,
+        discussion=args.discussion,
+    )
+    if args.require_workflow_labels and not label_assessment["valid"]:
+        print(
+            "Workflow label validation failed: "
+            + "; ".join(label_assessment["errors"]),
+            file=sys.stderr,
+        )
+        return 1
+
     user, password = get_credentials(host)
     if not user or not password:
         print(f"Could not get credentials for {host} "
@@ -59,11 +88,34 @@ def main(argv=None):
 
     import base64
     auth = f"Basic {base64.b64encode(f'{user}:{password}'.encode()).decode()}"
-    url = f"{repo_api_url(host, org, repo)}/issues"
+    base = repo_api_url(host, org, repo)
+    url = f"{base}/issues"
 
     try:
+        label_ids = []
+        if requested_labels:
+            existing = api_request("GET", f"{base}/labels", auth)
+            by_name = {lb["name"]: lb["id"] for lb in existing}
+            missing = [name for name in requested_labels if name not in by_name]
+            if missing:
+                print(f"Missing labels: {missing}", file=sys.stderr)
+                return 1
+            label_ids = [by_name[name] for name in requested_labels]
         data = api_request("POST", url, auth, {"title": args.title, "body": body})
+        if label_ids:
+            api_request(
+                "PUT",
+                f"{base}/issues/{data.get('number')}/labels",
+                auth,
+                {"labels": label_ids},
+            )
         print(f"Issue #{data.get('number')}: {data.get('html_url')}")
+        if not label_assessment["valid"]:
+            print(
+                "Workflow label recommendation: "
+                + "; ".join(label_assessment["errors"]),
+                file=sys.stderr,
+            )
         return 0
     except RuntimeError as e:
         print(f"Error: {e}", file=sys.stderr)

docs/architecture/control-plane-db-substrate.md

@@ -0,0 +1,105 @@
+# Control-plane DB substrate (#613)
+
+**Status:** Implemented (SQLite single-writer MVP)
+
+**ADR:** [`mcp-allocator-control-plane-observability-adr.md`](mcp-allocator-control-plane-observability-adr.md)
+
+**Module:** `control_plane_db.py`
+
+## Architecture statement
+
+> **DB coordinates, Gitea records, Sentry/GlitchTip observe; the bridge is the only path that turns observations into Gitea work.**
+
+## What this ships
+
+| Capability | Notes |
+|------------|--------|
+| Schema | `sessions`, `work_items`, `leases`, `assignments`, `terminal_locks`, `events`, `incident_links` |
+| Atomic assign+lease | `ControlPlaneDB.assign_and_lease` — one `BEGIN IMMEDIATE` transaction |
+| Mutation gate | `require_valid_assignment` — live lease + allowed action + non-terminal work + non-stale head |
+| Heartbeat / release / expire | Lease lifecycle helpers |
+| Terminal-lock index | Routing signal for #600 (terminal path first) |
+| `incident_links` | Provider-neutral link model for #612 — **not** assignable work; scope keys NULL-safe |
+
+## Hard rules (enforced in code)
+
+1. Assignable `work_items.kind` ∈ {`issue`, `pr`} only — **never** raw Sentry/GlitchTip incidents.
+2. Two concurrent sessions cannot both receive an active assignment on the same open work item (second gets `wait`).
+3. Merged/closed work items return `no_safe_work` at assign time, and `require_valid_assignment` fails closed if the work item becomes terminal later.
+4. Assignments pin `expected_head_sha`; mutations fail closed if the work item head drifts.
+5. SQLite path is the **single-writer MVP** (`GITEA_CONTROL_PLANE_DB`, default under `~/.cache/gitea-tools/control-plane/`). Multi-session multi-host production requires **Postgres** or a **single allocator daemon** (ADR §6).
+6. `incident_links` optional scope fields are stored as empty strings (never NULL) so UNIQUE is canonical across minimal upserts.
+7. Legacy `incident_links` migration collapses NULL-scope duplicates **only** when Gitea targets **and** all meaningful observation metadata agree (fingerprint, status, event_count, permalink, timestamps, linked PRs, etc.). Conflicting metadata fails closed — no silent discard.
+
+## Dependency chain
+
+```text
+#613 control-plane DB (this)  →  #600 allocator API  →  #612 incident bridge
+```
+
+- **#600** must call this substrate (not file locks / comment-only leases alone) for completion.
+- **#612** must write `incident_links` here and create **Gitea issues**; the allocator assigns those issues, not raw incidents.
+
+## Allocator API (#600)
+
+Module: `allocator_service.py` · MCP tool: `gitea_allocate_next_work`
+
+- Workers call `gitea_allocate_next_work(apply=false|true)` instead of self-selecting work.
+- `apply=false` returns a dry-run selection (`outcome=preview`) with skip reasons.
+- `apply=true` reserves via `ControlPlaneDB.assign_and_lease` (atomic assignment+lease).
+- Coordination source is **always** the control-plane DB — not file locks or comment-only leases.
+- Routing follows ADR §5.3 (REQUEST_CHANGES → author, terminal path first, foreign lease → wait).
+- Raw monitoring incidents are never candidates.
+
+
+## Lease lifecycle API (#601)
+
+Module: `lease_lifecycle.py` · MCP tools: `gitea_*_workflow_lease(s)`
+
+Active control-plane leases are **first-class workflow state**:
+
+| Tool | Purpose |
+|------|---------|
+| `gitea_list_workflow_leases` | List active (or all) leases for a repo |
+| `gitea_inspect_workflow_lease` | Freshness + `safe_next_action` for one lease id |
+| `gitea_adopt_workflow_lease` | Owner-resume or sanctioned reclaim with provenance |
+| `gitea_release_workflow_lease` | Explicit owner release (audited) |
+| `gitea_expire_workflow_leases` | Deterministic expire of past-`expires_at` leases |
+| `gitea_abandon_workflow_lease` | Abandon with required proof |
+| `gitea_reclaim_expired_workflow_lease` | Expire + re-assign with provenance |
+
+### Hard rules
+
+1. Control-plane DB is the coordination authority — file locks / comment-only leases are not authoritative alone.
+2. Active foreign leases cannot be stolen; inspect returns `wait_foreign_active`.
+3. Abandon requires `(dead_process OR missing_worktree)` and `no_live_mutation_risk`; foreign abandon also needs `operator_authorized` or full dead+missing+no_open_pr proof.
+4. Adopt provenance always records `adopted_from_session_id`, `adopted_by_session_id`, work identity, optional head SHA, and worktree path.
+5. Reviewer→merger comment handoff (`gitea_adopt_merger_pr_lease`) is unchanged and remains the Gitea-thread durability path.
+
+```bash
+python3 -m pytest tests/test_lease_lifecycle.py tests/test_control_plane_db.py tests/test_allocator_service.py tests/test_merger_lease_adoption.py -q
+```
+
+## Incident bridge (#612)
+
+Module: `incident_bridge.py` · MCP tools: `gitea_observability_*`
+
+- Bridge converts Sentry/GlitchTip observations → **normal Gitea issues** + `incident_links`.
+- Phase-1: `gitea_observability_reconcile_incident(apply=false|true)` with JSON observation.
+- Dry-run performs **no** Gitea mutation and **no** DB write.
+- Apply reuses existing links or creates one Gitea issue; never invents `work_items.kind=incident`.
+- Config: `GITEA_OBSERVABILITY_PROJECTS_JSON` or `GITEA_OBSERVABILITY_PROJECTS_FILE`.
+- Provider tokens never appear in issue bodies, links, or tool results.
+- Allocator sees bridge work only after a Gitea issue exists.
+
+## Non-goals (intentionally deferred)
+
+- Full unsupervised watchdog auto-filing (prefer explicit reconcile first)
+- Assuming GlitchTip writeback API equals Sentry without verification
+- Gitea comment/label mirror writers for every assignment (optional later)
+
+## Tests
+
+```bash
+python3 -m pytest tests/test_control_plane_db.py -q
+```

docs/architecture/mcp-allocator-control-plane-observability-adr.md

@@ -0,0 +1,301 @@
+# ADR: MCP allocator, control-plane DB, and Sentry/GlitchTip incident bridge architecture
+
+- **Status:** Accepted (implementation pending; blocks code for #600 / #612 / #613)
+- **Date:** 2026-07-09
+- **Tracking issues:**
+  - [#613](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/613) — control-plane DB (first)
+  - [#600](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/600) — allocator API (second)
+  - [#612](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/612) — Sentry/GlitchTip incident bridge (third)
+- **Related:** existing GlitchTip contracts (`glitchtip-to-gitea-workflow-design.md`, `glitchtip-gitea-deduplication-linking-design.md`), trust boundaries (`tool-boundaries.md`, `safety-model.md`), current leases (`pr_work_lease.py`, `issue_lock_store.py`)
+
+## 1. Context
+
+Multiple LLM sessions can independently inspect the Gitea queue and start the same issue or PR. Existing coordination (Gitea comment leases, local issue-lock files, labels) often detects collisions **after** work has started. Parallel issues #600, #612, and #613 each describe part of a fix; without one ADR they risk overlapping stores, wrong dependency order, and trust-boundary violations.
+
+This ADR is the canonical architecture decision **before** implementing those issues.
+
+## 2. Decision summary (core)
+
+| Layer | Owns | Must not |
+|-------|------|----------|
+| **Control-plane DB** | Sessions, atomic assignment, leases, heartbeats, terminal-lock index, events, incident links | Bypass Gitea workflow gates or replace issue/PR history |
+| **Gitea** | Durable work record: issues, PRs, comments, labels, reviews, merges | Be the only concurrency lock under multi-session load |
+| **Sentry / GlitchTip** | Incidents, events, provider UI | Assign work, approve/merge/close, or mutate Gitea outside the bridge |
+| **Incident bridge** | Provider adapters, reconcile/create Gitea issues, link storage upsert, optional provider writeback | Hand raw provider incidents to the allocator as work items |
+
+**One-liner:** **DB coordinates. Gitea records. Sentry/GlitchTip observe. The bridge is the only path that turns observations into Gitea work. The allocator assigns only Gitea issues/PRs, never raw monitoring incidents.**
+
+## 3. Dependency order
+
+Implementation **must** follow:
+
+```text
+#613 control-plane DB  →  #600 allocator API  →  #612 incident bridge
+   (atomic substrate)      (routing policy)       (feeds Gitea work)
+```
+
+| Issue | Role | Depends on |
+|-------|------|------------|
+| **#613** | Durable coordination DB + lease/assignment transactions | — |
+| **#600** | `gitea_allocate_next_work` policy and tool surface | **#613** (hard) |
+| **#612** | Multi-project Sentry/GlitchTip → Gitea bridge | **#613** for `incident_links` index; **#600** before treating bridge-created issues as allocator feed in multi-worker prod |
+
+**Hard rules:**
+
+1. Do **not** implement #600 on file locks / comment-only leases and call it done.
+2. Do **not** ship #612 as a second assignment system for raw incidents.
+3. Partial previews (read-only list tools, schema stubs) may land earlier if they do not claim “allocator complete” or “bridge complete.”
+
+## 4. Authority boundaries
+
+### 4.1 Gitea (durable source of truth for work)
+
+Gitea remains authoritative for:
+
+- Issue and PR identity, titles, bodies, state (open/closed/merged)
+- Comments, reviews, approvals, REQUEST_CHANGES
+- Labels and workflow status labels (`status:ready`, `status:in-progress`, …)
+- Merges, closes, branch refs as recorded by Gitea/git hosting
+
+Operators and auditors read **history** from Gitea.
+
+### 4.2 Control-plane DB (coordination / index)
+
+The control-plane DB is authoritative for **live multi-session coordination**:
+
+- Which session holds which assignment/lease
+- Heartbeat freshness and expiry
+- Terminal-lock index for routing
+- Event log of allocation/lease transitions
+- `incident_links` index (see §8)
+
+It is **not** a substitute for Gitea history. When DB and Gitea disagree on durable work state (e.g. PR already merged), **Gitea wins**; the DB is reconciled.
+
+### 4.3 Sentry / GlitchTip (observe)
+
+Providers own incident lifecycle and raw event data. They:
+
+- Must **not** bypass Gitea gates
+- Must **not** assign LLM work
+- May receive **writeback** of resolution status only through the bridge, when configured and supported
+
+### 4.4 Trust boundaries for credentials
+
+Aligned with `docs/tool-boundaries.md` and `docs/safety-model.md`:
+
+- **One MCP server process per trust boundary** remains the default.
+- Monitor API tokens (Sentry/GlitchTip) live in the **bridge/orchestrator boundary** (or a dedicated observability write/read profile), **not** blindly inside every Gitea MCP author/reviewer/merger process.
+- Gitea tokens stay on Gitea MCP profiles only.
+- Orchestrators compose services; they must not become a single credential pool that silently mixes Gitea write + monitor tokens into every worker.
+
+Tool names such as `gitea_observability_*` may exist as a **namespace façade** only if the runtime still enforces separate credential scopes (e.g. bridge process vs pure Gitea mutation process).
+
+## 5. Allocator rules (#600 on top of #613)
+
+### 5.1 Worker contract
+
+- Workers **do not self-select** work under the standard multi-LLM workflow.
+- Workers call **`gitea_allocate_next_work`** (with `apply=true` only when taking work).
+- Mutations that claim exclusive work require a **valid assignment and lease** from the control-plane DB.
+- Return values include at least: role, issue/PR number, expected head SHA (when applicable), lease ID, expiry, allowed actions, forbidden actions — or a non-assignment outcome (`WAIT`, terminal-path block, no safe work, needs controller, etc.).
+
+### 5.2 Atomic reserve
+
+- **Assignment creation and lease creation happen in one DB transaction.**
+- Two concurrent sessions **must not** receive the same issue/PR as assigned work.
+- On contention: second session gets **WAIT** or **owner-resume**, never a duplicate lease.
+
+### 5.3 Routing policy
+
+| Condition | Route |
+|-----------|--------|
+| Current-head **REQUEST_CHANGES** | **Author** (not reviewer/merger) |
+| **Stale** approval (approval not on current head) | **Reviewer** |
+| **Clean** approval on current head, mergeable | **Merger** |
+| Contested / contaminated approval | **Diagnosis / reconciler** (controller path) |
+| Active **foreign** lease | **WAIT** or **owner-resume** |
+| **Terminal-review** lock present | **Terminal-path resolution first** before downstream review work |
+| Merged / closed PR | **Never assign** |
+| Issue locked by sanctioned lease | **Never assign** to another worker unless lease cleanup/adoption is sanctioned |
+
+### 5.4 Relationship to Gitea mirrors
+
+After a successful assignment, the allocator (or sanctioned tooling) may update Gitea comments/labels so humans and audits see state. Those mirrors **are not** the sole lock source.
+
+## 6. Control-plane DB topology (#613)
+
+### 6.1 Preferred architecture (multi-daemon / Proxmox)
+
+For true multi-session concurrency (multiple MCP daemons, multiple hosts, or Proxmox VMs):
+
+**Prefer either:**
+
+1. **Shared Postgres** used by all Gitea MCP profiles / allocator callers, **or**
+2. A **single allocator daemon** (sole writer to the coordination store) that all workers call over MCP/RPC.
+
+Both satisfy: one transactional authority for “who has this work item.”
+
+### 6.2 SQLite MVP
+
+SQLite is acceptable **only** for a **single-writer MVP**:
+
+- One process owns writes (allocator daemon or single co-located MCP server), **or**
+- Documented single-host single-daemon experiments with file locking and no multi-host claims.
+
+SQLite is **not** sufficient to declare multi-session production readiness when four independent LLM sessions talk to four MCP processes without a shared writer.
+
+### 6.3 Suggested entities (normative shape)
+
+Minimum logical entities (names may vary; semantics must not):
+
+1. **sessions** — session_id, role/profile, namespace, pid, started_at, last_heartbeat_at, status  
+2. **work_items** — remote/org/repo, kind ∈ {`issue`, `pr`}, number, state, priority, current_head_sha, updated_at  
+3. **leases** — lease_id, work_item_id, session_id, role, phase, expires_at, heartbeat_at, status  
+4. **assignments** — assignment_id, work_item_id, session_id, allowed_action(s), expected_head_sha, status  
+5. **terminal_locks** — repo/org, terminal_pr, review_id, decision, status, cleanup_state  
+6. **events** — event_id, work_item_id, type, message, created_at  
+7. **incident_links** — provider-neutral link model (see §8)
+
+**Explicit non-kind:** do **not** use `work_items.kind = sentry_incident` (or glitchtip_incident) as an assignable work unit. Incidents become work only after the bridge creates/links a **Gitea issue**.
+
+## 7. Lease migration (avoid split-brain)
+
+### 7.1 Target state
+
+- **Primary** for live coordination: **control-plane DB** leases and assignments.
+- **Gitea comment leases** (e.g. `<!-- mcp-review-lease:v1 -->`) and **local issue-lock files** become:
+  - **mirrors** of DB state for human visibility / recovery, and/or
+  - written **only** through the allocator (or sanctioned lease tools that update DB + mirror in one workflow).
+
+### 7.2 Migration rules
+
+1. New exclusive claims go through DB assignment+lease first.
+2. Comment lease / file lock writers that skip the DB are **deprecated** once #613+#600 land.
+3. Reconciler tooling heals: expired DB lease, orphan Gitea comment, orphan local file — fail closed when ambiguous.
+4. **Split-brain is a defect:** “DB free + Gitea comment leased” or “DB leased + Gitea free” must be detectable and reconcilable; production paths must not rely on two independent writers.
+
+### 7.3 Heartbeats
+
+- Heartbeats update the **DB**.
+- Optional Gitea summaries must avoid comment spam (periodic summary or edit-in-place policy).
+
+## 8. Observability bridge (#612)
+
+### 8.1 Role
+
+The bridge:
+
+1. Scans configured Sentry and/or GlitchTip projects (multi-project mappings).
+2. Deduplicates against existing links / Gitea issues.
+3. Creates or updates **normal Gitea issues** (labels, sanitized body, provider URL).
+4. Upserts **one** canonical `incident_links` row.
+5. Optionally writes resolution status back to the provider when supported.
+6. Never approves, merges, closes, or otherwise bypasses Gitea workflow gates.
+
+### 8.2 Allocator interaction
+
+- The allocator sees observability work **only after** a Gitea issue exists (typically `status:ready` + observability labels).
+- **Do not assign raw Sentry/GlitchTip incidents** as work items.
+- Bridge AC “allocator can see linked issues” means: **as ordinary Gitea issues**, not a special incident queue.
+
+### 8.3 Provider adapters and trust
+
+- Separate **Sentry** and **GlitchTip** adapters; document API differences; do not assume writeback parity.
+- Self-hosted base URLs supported (e.g. `https://sentry.prgs.cc`).
+- Tokens from environment / secret store only.
+- Prefer phase-1 **explicit reconcile / dry-run** before unsupervised watchdog auto-filing, consistent with existing GlitchTip filing safety contracts.
+
+### 8.4 Home of implementation
+
+Filing/orchestration may live in:
+
+- a control-plane / bridge package or profile, and/or  
+- Gitea-Tools as operator-facing tools that **delegate** to that boundary,
+
+but must not violate §4.4 credential isolation.
+
+## 9. Incident link storage (single source of truth)
+
+### 9.1 Canonical model: `incident_links` (provider-neutral)
+
+Prefer a **provider-neutral** model over Sentry-only `sentry_links`:
+
+| Field (logical) | Purpose |
+|-----------------|---------|
+| provider | `sentry` \| `glitchtip` \| … |
+| provider_base_url | Self-hosted base |
+| provider_org / provider_project | Mapping key |
+| provider_issue_id / provider_short_id | Provider identity |
+| provider_permalink | Human URL |
+| fingerprint | Stable dedupe key when available |
+| gitea_org / gitea_repo / gitea_issue_number | Linked work |
+| linked_pr_numbers | Optional PR association |
+| first_seen / last_seen / event_count | Summary metrics (safe) |
+| status | link lifecycle |
+| release_resolved_at / last_sync_at | Sync bookkeeping |
+
+### 9.2 Gitea as human mirror
+
+- Issue body markers / structured comments (e.g. HTML comment metadata) remain the **human- and audit-visible mirror**.
+- They must stay consistent with `incident_links` but are **not** a second independent write path for inventing links without the bridge.
+
+### 9.3 One truth
+
+- **Do not** maintain two independent systems of record (e.g. full mapping only in #612 storage **and** a separate #613 `sentry_links` with different semantics).
+- #612 implementation **must** use the same `incident_links` model introduced under #613 (or a single agreed store both reference).
+
+## 10. Security and redaction
+
+Mandatory for bridge payloads, Gitea issue bodies/comments, DB-stored event summaries, and logs:
+
+- No API tokens, DSNs, passwords, cookies, auth headers
+- No keychain IDs, private config contents, raw session-state files, full prompt bodies
+- Sanitize stack locals and request data; store only safe tags/context
+- Logs must never print provider tokens or DSNs
+- Redaction tests are required for #612
+
+## 11. Non-goals
+
+- Replace Gitea as the durable workflow record  
+- Let Sentry/GlitchTip assign work or mutate Gitea workflow state outside sanctioned tools  
+- Let the bridge approve, merge, close, release, or bypass Gitea gates  
+- Implement #600 on file locks / comments alone and call allocator complete  
+- Treat raw monitoring incidents as `work_items` for the allocator  
+- Put monitor tokens into every Gitea MCP worker process by default  
+
+## 12. Consequences
+
+### Positive
+
+- Clear implementation order and ownership  
+- Multi-session safety becomes testable at the DB layer  
+- Observability becomes assignable work without special-casing the allocator  
+- Trust boundaries stay compatible with existing control-plane docs  
+
+### Costs / risks
+
+- Migration from comment/file leases requires reconciler work  
+- Shared Postgres or single allocator daemon is operational cost  
+- Bridge must be careful not to spam Gitea issues (dedupe, caps, policy modes)  
+
+### Follow-ups (documentation only until implemented)
+
+1. Update #600, #612, and #613 bodies to **link this ADR** and restate hard dependencies.  
+2. Implement #613 schema + atomic assign/lease.  
+3. Implement #600 against the DB.  
+4. Implement #612 against `incident_links` + Gitea create/update.  
+5. Deprecate dual writers for leases.
+
+## 13. Acceptance criteria for *this* ADR
+
+This document is accepted when:
+
+1. It is merged into `docs/architecture/` on the default branch.  
+2. #600 / #612 / #613 (or their PRs) reference this path as the architecture source of truth.  
+3. No implementation PR for those issues claims completion without conforming to §§2–11.
+
+## 14. Document history
+
+| Date | Change |
+|------|--------|
+| 2026-07-09 | Initial ADR: dependency order, authority model, topology, lease migration, bridge, `incident_links`, security, non-goals |

docs/architecture/mcp-stable-control-runtime-policy-adr.md

@@ -0,0 +1,198 @@
+# ADR: Stable control runtime vs dev runtime (Gitea MCP)
+
+- **Status:** Accepted (policy effective immediately for LLM sessions; tooling may lag)
+- **Date:** 2026-07-09
+- **Tracking issue:** [#615](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/615)
+- **Related:**
+  - [#543](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/543) / `docs/mcp-namespace-health.md` — client-namespace health
+  - `docs/mcp-namespace-eof-recovery.md` — reconnect-only EOF recovery (no PID kill)
+  - [#558](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/558) / `docs/mcp-daemon-import-guard.md` — sanctioned daemon
+  - [#557](https://gitea.prgs.cc/Scaled-Tech-Consulting/Gitea-Tools/issues/557) / `docs/bootstrap-review-path.md` — controller bootstrap for self-hosted fixes
+  - Allocator / control-plane ADR: `docs/architecture/mcp-allocator-control-plane-observability-adr.md` (#613 / PR #614)
+
+## 1. Context
+
+The Gitea MCP server is the **control plane** for real issue/PR mutations (create, comment, lock, review, merge, etc.). When author/reviewer/merger/reconciler sessions kill or restart that process, relaunch it from a feature worktree, or edit the checkout that process loads, operators observe:
+
+- Mid-session identity/preflight resets
+- Stale-runtime vs master parity failures
+- IDE transport EOF / “tool not found” while code on disk has changed
+- Accidental production mutations from experimental code
+
+This ADR separates **stable control runtime** from **dev/test runtime** and defines promotion proof.
+
+## 2. Decision
+
+### 2.1 Stable control runtime
+
+The Gitea MCP server used for **real workflow mutations** is the **stable control runtime**.
+
+Characteristics:
+
+- Loads a known, promoted revision of Gitea-Tools (or the packaged release layout operators designate)
+- Registered in the IDE/client as the production namespaces (`gitea-tools`, `gitea-reviewer`, `gitea-merger`, `gitea-reconciler`, etc.)
+- Holds production profile credentials via sanctioned keychain/env paths only
+
+### 2.2 Dev / test runtime
+
+MCP **server code** development and testing:
+
+- Happens in isolated **`branches/`** worktrees (or other non-stable checkouts)
+- May use a **separate** dev/test MCP runtime/process when process-level testing is required
+- **Must not** be used for real Gitea mutations on production issues/PRs
+
+### 2.3 Forbidden actions (normal sessions)
+
+Normal **author, reviewer, merger, and reconciler** LLM sessions **must not**:
+
+| Forbidden | Why |
+|-----------|-----|
+| Kill the running MCP server process | Drops all concurrent sessions; loses preflight state |
+| Restart / relaunch the MCP server process | Same as kill; causes stale/identity churn mid-workflow |
+| Relaunch MCP from a development worktree | Runs unpromoted code against production mutations |
+| Edit files in the stable runtime checkout | Hot-mutates control plane under concurrent users |
+| Use experimental/dev MCP for real Gitea mutations | Bypasses promotion proof and audit expectations |
+| Bypass or self-reset a stale master-parity gate | The gate is fail-closed; only an operator reload restores parity |
+
+**LLM-allowed vs operator-owned (authoritative split):**
+
+| Actor | May do | Must not do |
+|-------|--------|-------------|
+| **LLM session** | Call tools on the already-running stable namespaces; **client reconnect** after transport EOF (no process kill); pass `worktree_path` / role worktree args; report blockers and stop mutations when unhealthy/stale | Kill, restart, or relaunch any MCP process; bump config mtimes to force reload; edit the stable checkout; switch to a dev MCP for production mutations |
+| **Operator / release-manager** | Supervised restart/reload of the **stable** control runtime; dual-namespace client configuration; §2.4 promotions; incident recovery | — |
+
+EOF / transport recovery for LLM sessions: **client reconnect only** (see `docs/mcp-namespace-eof-recovery.md`). Do not “fix” health by killing PIDs or bumping MCP config mtimes as a normal session procedure.
+
+This ADR **supersedes** any older runbook wording that told the LLM to relaunch or restart the client/MCP as a self-service step. Where `docs/llm-workflow-runbooks.md` (or wiki runbooks) discuss dual-namespace setup or workspace rebind, **process restart/relaunch is operator-owned**; the LLM stops, reports, and waits.
+
+### 2.4 Promotion (operator / release-manager only)
+
+Promotion of a new revision into the stable control runtime is an **explicit operator/release-manager action**, not an LLM self-service step.
+
+A promotion **must record** (issue comment, release note, or promotion ledger):
+
+| Field | Description |
+|-------|-------------|
+| **previous runtime SHA** | Commit previously loaded by stable runtime |
+| **promoted runtime SHA** | Commit after promotion |
+| **source branch/PR** | Where the change was reviewed |
+| **restart/reload method** | How the process was cycled (e.g. supervised restart, client reload) |
+| **health check proof** | Client-namespace probe success (`gitea_whoami` / namespace health) |
+| **identity/profile proof** | Expected profile(s) and username(s) after reload |
+| **workspace/root proof** | Stable checkout path / root matches intended layout |
+| **mutation capability proof** | Required permissions for the target role present; forbidden ops still forbidden |
+| **rollback instructions** | How to restore previous SHA and re-verify health |
+
+Suggested durable marker:
+
+```text
+## MCP STABLE RUNTIME PROMOTION (#615)
+
+Status: COMPLETED | ROLLED_BACK | ABORTED
+Previous-SHA: <full sha>
+Promoted-SHA: <full sha>
+Source-PR: <number>
+Source-Branch: <name>
+Reload-Method: <text>
+Health-Proof: client_namespace whoami OK / assess_mcp_namespace_health OK
+Identity-Proof: profile=<name> user=<name>
+Workspace-Proof: root=<path>
+Mutation-Proof: allowed_ops include <…>; forbidden include <…>
+Rollback: checkout <previous sha>; reload method <…>; re-run health/identity proofs
+Operator: <username>
+Timestamp: <ISO-8601>
+```
+
+### 2.5 Unhealthy stable runtime → stop work
+
+If the stable MCP runtime is **unhealthy**, including any of:
+
+- client-namespace probes fail
+- wrong identity / wrong profile
+- wrong workspace root
+- missing mutation capability for the intended role
+- persistent EOF after **client reconnect**
+- **master parity is stale** (`startup_head` behind on-disk `master` / `restart_required` from the parity gate)
+
+then:
+
+1. **Normal PR / review / merge / issue-mutation work must stop immediately.**
+2. The session **reports** the unhealthy/stale state (tool error, CTH, or operator handoff) with startup vs current head when known.
+3. Do **not** improvise: no LLM process kill/restart, no dev-worktree MCP for production mutations, no env escape hatches, no manual gate bypass.
+4. Resume only after:
+   - an **operator** restores the runtime (see §2.6 for routine post-merge parity reload), **or**
+   - a **controlled** promotion/rollback completes with the §2.4 promotion record, **or**
+   - a controller invokes the narrow **bootstrap review path** (#557) when the defect is self-hosted and documented,
+   - **and** the session re-verifies health (and master parity when applicable) before the next mutation.
+
+### 2.6 Routine post-merge master-parity staleness (operator reload, not promotion)
+
+**Symptom:** After merges land on `master`, a long-lived stable MCP process still runs the pre-merge `startup_head`. The master-parity gate marks the server **stale** / `restart_required` and **blocks mutations**.
+
+**Sanctioned response (authoritative):**
+
+| Step | Actor | Action |
+|------|-------|--------|
+| 1 | LLM session | Mutations stop immediately when the gate reports stale. |
+| 2 | LLM session | Report the stale state (startup head, current master head, that operator reload is required). Do not retry mutations. |
+| 3 | **Operator** | Reload/restart the **stable** control MCP so it loads current `master` (supervised client/daemon reload). |
+| 4 | LLM session | Resume only after startup/current-head parity is verified (e.g. `gitea_get_runtime_context` / parity assessment shows in parity). |
+
+**Not a §2.4 promotion:** Catching the already-designated stable control checkout up to a newly advanced `master` is a **routine operator reload** of the stable runtime. It does **not** require the nine-field promotion ledger. Use §2.4 only when **changing which unpromoted/dev revision** becomes the stable control runtime (new source branch/PR into the stable designation).
+
+**Code note:** `master_parity_gate.py` may still say “restart the server” in machine-facing reason strings. That string names the **operator recovery action**, not an LLM self-service instruction. This ADR and the runbooks define the actor split.
+
+## 3. Relationship to other controls
+
+| Doc / mechanism | Interaction |
+|-----------------|-------------|
+| Namespace health (#543) | Proves IDE client can call tools; does not authorize restart |
+| EOF recovery | Reconnect only; no process kill |
+| Daemon import guard (#558) | Mutations require sanctioned daemon; not a bare shell import |
+| Bootstrap path (#557) | Only controller-authorized exception when live runtime cannot review its own fix |
+| Allocator / control-plane ADR | Coordination DB is separate; still depends on a healthy MCP surface for Gitea writes |
+
+## 4. Consequences
+
+### Positive
+
+- Predictable control plane for concurrent LLMs
+- Clear operator-only promotion gate with rollback
+- Aligns session behavior with health/EOF docs already landed
+
+### Costs
+
+- LLM sessions must wait when runtime is sick (no DIY restart)
+- Operators must maintain promotion discipline and dual-runtime config if they use a dev MCP
+
+### Non-goals
+
+- Does not ban operator-supervised restarts during incidents
+- Does not replace CI or code review for MCP changes
+- Does not authorize editing stable checkout “because tests need a quick fix”
+
+## 5. Implementation follow-ups (optional tooling)
+
+These may land in later issues; the **policy binds sessions now**:
+
+1. Session preflight that refuses mutations if workspace root equals a `branches/` feature worktree configured as “dev only.”
+2. Explicit `runtime_kind=stable|dev` in MCP config and `gitea_whoami` profile metadata.
+3. Promotion checklist script that emits the durable promotion marker fields.
+
+**Not optional (issue #615 acceptance criterion 2):** operator guide and runbooks **must** cross-link this ADR (see §6). Cross-links are documentation acceptance, not deferred tooling.
+
+## 6. Acceptance for this ADR
+
+1. Document merged under `docs/architecture/mcp-stable-control-runtime-policy-adr.md`.
+2. **Operator guide / runbooks cross-link this ADR** (`docs/wiki/Operator-Guide.md`, `docs/wiki/Runbooks.md`, `docs/llm-workflow-runbooks.md`).
+3. Issue #615 references this path.
+4. LLM/operator runbooks treat kill/restart/relaunch-from-worktree as **LLM violations**; process restart is **operator-owned**.
+5. Unhealthy runtime (including **stale master parity**) stops normal mutation work until operator restore/reload, promotion/rollback, or #557 bootstrap — then re-verify parity before mutating.
+6. Routine post-merge parity reload is documented as operator reload (§2.6), not an LLM self-restart and not a full §2.4 promotion.
+
+## 7. Document history
+
+| Date | Change |
+|------|--------|
+| 2026-07-09 | Initial ADR: stable vs dev runtime, forbidden session actions, promotion proof fields, stop-work rule |
+| 2026-07-16 | Review 443 remediation (#615 / PR #616): mandatory cross-links; LLM vs operator restart split; routine post-merge parity staleness (§2.6); stale parity in §2.5 unhealthy triggers |

docs/canonical-state-comments.md

@@ -0,0 +1,183 @@
+# Canonical State Comments
+
+Gitea is the durable system of record for workflow continuation. When a
+comment changes issue, PR, or discussion state, it should leave enough
+information for the next role to continue without private chat history.
+
+Canonical comments answer:
+
+- what state the object is in
+- who acts next
+- what the next actor should do
+- the exact prompt the next actor should run
+- which proof, blocker, or dependency matters
+
+Non-workflow discussion comments do not need this template.
+
+## Issue State
+
+Use this when an issue becomes ready, blocked, in progress, PR-open,
+superseded, merged, or otherwise changes workflow direction.
+
+```text
+## Canonical Issue State
+
+STATE:
+<ready-for-author | in-progress | blocked | PR-open | needs-review | ready-to-merge | merged | closed | superseded>
+
+WHO_IS_NEXT:
+<controller | author | reviewer | merger | reconciler | user>
+
+NEXT_ACTION:
+<specific one-sentence action>
+
+NEXT_PROMPT:
+<paste-ready prompt for the next role>
+
+WHAT_HAPPENED:
+<latest meaningful event>
+
+WHY:
+<decision rationale>
+
+RELATED_DISCUSSION:
+<link/reference or none>
+
+RELATED_PRS:
+- #...
+
+BRANCH:
+<branch or none>
+
+HEAD_SHA:
+<40-character SHA or none>
+
+VALIDATION:
+<tests/proofs or none>
+
+BLOCKERS:
+<blocker and unblock condition, or none>
+
+LAST_UPDATED_BY:
+<identity/profile/date>
+```
+
+## PR State
+
+Use this when a PR needs review, receives changes requested, is approved,
+is stale, is superseded, or becomes ready for merge.
+
+```text
+## Canonical PR State
+
+STATE:
+<needs-review | changes-requested | approved | stale-approval | ready-to-merge | merged | blocked | superseded>
+
+WHO_IS_NEXT:
+<controller | author | reviewer | merger | reconciler | user>
+
+NEXT_ACTION:
+<specific one-sentence action>
+
+NEXT_PROMPT:
+<paste-ready prompt for the next role>
+
+WHAT_HAPPENED:
+<latest meaningful event>
+
+WHY:
+<decision rationale>
+
+ISSUE:
+#...
+
+BASE:
+<branch>
+
+HEAD:
+<branch>
+
+HEAD_SHA:
+<40-character SHA>
+
+REVIEW_STATUS:
+<none | approved | changes-requested | stale | contaminated>
+
+VALIDATION:
+<tests/proofs>
+
+BLOCKERS:
+<blockers or none>
+
+SUPERSEDES:
+<PRs or none>
+
+SUPERSEDED_BY:
+<PR or none>
+
+MERGE_READY:
+<yes/no and why>
+
+LAST_UPDATED_BY:
+<identity/profile/date>
+```
+
+## Discussion Summary
+
+Discussions should normally have at least five substantive comments before
+conversion into issues. A controller may waive that only for tiny mechanical,
+urgent, or explicitly trivial work.
+
+```text
+## Canonical Discussion Summary
+
+STATE:
+<needs-more-discussion | ready-for-issues | issues-created | closed>
+
+WHO_IS_NEXT:
+<controller | author | reviewer | user>
+
+DECISION:
+<what was decided>
+
+WHY:
+<reasoning and tradeoffs>
+
+SUBSTANTIVE_COMMENTS:
+<count and summary>
+
+ISSUES_TO_CREATE_OR_CREATED:
+- #...
+
+DEPENDENCY_ORDER:
+<order or none>
+
+NON_GOALS:
+<non-goals>
+
+OPEN_QUESTIONS:
+<questions or none>
+
+NEXT_ACTION:
+<specific one-sentence action>
+
+NEXT_PROMPT:
+<paste-ready prompt for the next role>
+
+LAST_UPDATED_BY:
+<identity/profile/date>
+```
+
+## Validation Rules
+
+The final-report validator rejects canonical state update claims when the
+report omits the canonical block or when the block lacks:
+
+- `STATE`
+- `WHO_IS_NEXT`
+- `NEXT_ACTION`
+- `NEXT_PROMPT`
+
+It also rejects vague next actions such as `continue`, ready-to-merge states
+without approval/head-SHA proof, superseded states without canonical item
+proof, and blocked states without an unblock condition.

docs/canonical-thread-handoff.md

@@ -0,0 +1,142 @@
+# Canonical Thread Handoff (CTH)
+
+**CTH** = **Canonical Thread Handoff**
+
+A CTH comment is the authoritative workflow handoff in a Gitea issue or PR
+thread. It records current state, decisions, blockers, proof, and the exact
+next prompt/action for the next LLM or person.
+
+CTH comments complement — but do not replace — formal Gitea review state.
+A CTH may summarize an APPROVE or REQUEST_CHANGES decision, yet merge gates
+still require the live Gitea review verdict.
+
+## CTH comment types
+
+- `CTH: State Handoff`
+- `CTH: Controller Decision`
+- `CTH: Author Handoff`
+- `CTH: Reviewer Handoff`
+- `CTH: Merger Handoff`
+- `CTH: Supersession Notice`
+- `CTH: Blocker`
+
+## Required base template
+
+```md
+## CTH: <Type>
+
+Status:
+Next owner:
+Current blocker:
+Decision:
+Proof:
+Next action:
+Ready-to-paste prompt:
+```
+
+Extended fields may map to canonical issue/PR state templates from #495 when
+that layer is available. Until then, keep the base fields complete.
+
+## Discovery and posting rules
+
+Before acting in an issue or PR thread:
+
+1. **Find the latest CTH comment** before doing work.
+2. Treat the **latest valid CTH** as the current handoff state.
+3. **Post a new CTH** when you finish, block, skip, supersede, request
+   changes, approve, or hand off.
+4. Do **not** rely on stale non-CTH comments when a newer CTH exists.
+5. Casual discussion comments do not need to be CTH comments.
+
+## Examples
+
+### PR approved and ready for merger
+
+```md
+## CTH: Reviewer Handoff
+
+Status: approved_at_current_head
+Next owner: merger
+Current blocker: none
+Decision: APPROVE recorded at head abc123...
+Proof: gitea_submit_pr_review performed; visible verdict APPROVE
+Next action: eligible merger merges PR #N with pinned expected_head_sha
+Ready-to-paste prompt: Merge PR #N for issue #M if live head still abc123... and merge gates pass.
+```
+
+### PR request-changes back to author
+
+```md
+## CTH: Reviewer Handoff
+
+Status: request_changes_at_current_head
+Next owner: author
+Current blocker: unresolved findings in validation report
+Decision: REQUEST_CHANGES at head def456...
+Proof: gitea_submit_pr_review performed; blocking review visible
+Next action: author fixes findings and pushes; reviewer re-validates fresh head
+Ready-to-paste prompt: Fix PR #N review findings, push to feat/issue-M-..., post Author Handoff CTH.
+```
+
+### Duplicate / superseded PR closure
+
+```md
+## CTH: Supersession Notice
+
+Status: superseded
+Next owner: controller
+Current blocker: duplicate branch/PR work
+Decision: close PR #N; continue on PR #M
+Proof: duplicate gate linked open PR #M for issue #K
+Next action: controller closes superseded PR and records canonical state
+Ready-to-paste prompt: Close superseded PR #N; confirm PR #M remains canonical for issue #K.
+```
+
+### Blocked workflow due to dirty root/worktree
+
+```md
+## CTH: Blocker
+
+Status: blocked_preflight
+Next owner: operator
+Current blocker: dirty control checkout / branches worktree mismatch
+Decision: stop before mutation
+Proof: verify_preflight_purity failed; workspace diagnostics attached
+Next action: repair worktree or relaunch MCP from branches/ worktree
+Ready-to-paste prompt: Repair dirty workspace, relaunch MCP from branches/review-prN, retry review.
+```
+
+### Stale head requiring fresh review
+
+```md
+## CTH: Reviewer Handoff
+
+Status: stale_head
+Next owner: reviewer
+Current blocker: live head differs from pinned review head
+Decision: prior approval not valid for current head
+Proof: expected_head_sha mismatch at merge gate
+Next action: re-run validation and post fresh review decision
+Ready-to-paste prompt: Re-validate PR #N at live head, dry-run review gates, submit fresh verdict.
+```
+
+### Issue implementation handoff
+
+```md
+## CTH: Author Handoff
+
+Status: implementation_complete_pending_review
+Next owner: reviewer
+Current blocker: none
+Decision: PR #N ready for review at head fedcba...
+Proof: tests passed in branches/issue-M-...; PR opened with Closes #M
+Next action: reviewer acquires lease and validates PR #N
+Ready-to-paste prompt: Review PR #N for issue #M; pin head fedcba... before mutations.
+```
+
+## Related
+
+- [`llm-workflow-runbooks.md`](llm-workflow-runbooks.md)
+- [`../skills/llm-project-workflow/templates/canonical-thread-handoff.md`](../skills/llm-project-workflow/templates/canonical-thread-handoff.md)
+- #495 — canonical next-action comment fields
+- #496 — fail-closed validation for workflow-changing comments

docs/developer-testing-guidelines.md

@@ -54,6 +54,25 @@ Use `-q` for a compact summary and `-v` to see individual test names.
 ./venv/bin/python -m pytest tests/ -q
 ```
 
+### Web UI suite (#436)
+
+Hermetic unittest modules matching `test_webui_*.py` cover route rendering,
+read-only guards, registry/prompt/queue loaders, and optional child-issue
+modules when present on the branch.
+
+```bash
+./scripts/test-webui
+./scripts/ci-webui-check          # skip unless the diff touches web UI paths
+WEBUI_CI_FORCE=1 ./scripts/ci-webui-check
+```
+
+`scripts/test-webui` defaults `WEBUI_TEST_OFFLINE=1` so route coverage never
+needs Gitea credentials or MCP daemon credential access. Set
+`WEBUI_TEST_OFFLINE=0` only for explicit operator live-fetch checks.
+
+Wire `scripts/ci-webui-check` into Jenkins (or equivalent) for PRs that touch
+`webui/`, `tests/test_webui_*`, or `docs/webui*`.
+
 ### Run targeted tests
 
 ```bash

docs/examples/two-comment-workflow-examples.md

@@ -0,0 +1,43 @@
+# Two-comment workflow examples (#507)
+
+Paired `[CONTROLLER HANDOFF]` + `[THREAD STATE LEDGER]` comments for Gitea threads.
+See `thread_state_ledger_examples.py` for machine-checked fixtures.
+
+## Approved review posted
+
+**Handoff** (detailed): identity, worktree, validation commands, mutation ledger with
+`gitea_submit_pr_review → APPROVED review posted to Gitea`.
+
+**Ledger** (concise):
+
+```markdown
+[THREAD STATE LEDGER] PR #487 — APPROVED review posted to Gitea
+
+What is true now:
+- PR state: open
+- Server-side decision state: APPROVED review posted to Gitea
+- Local verdict/state: APPROVE verdict prepared locally
+
+What is blocked:
+- Blocker classification: no blocker
+
+Who/what acts next:
+- Next actor: merger
+- Required action: merge on explicit operator command
+- Do not do: re-post APPROVE
+```
+
+## Approve validated locally but blocked before posting
+
+Ledger must show `no server-side state changed` under server-side decision state and
+`APPROVE verdict prepared locally` under local verdict/state.
+
+## Environment / tooling blocker
+
+Ledger blocker classification: `environment/tooling blocker`. Mutation ledger:
+`none — no server-side state changed`.
+
+## Stale head blocker
+
+Ledger: `approval_at_current_head is false`; classification `stale head`.
+Do not do: merge with stale approval.

docs/gitea-dual-namespace-deployment.md

@@ -22,9 +22,12 @@ launched with exactly one static execution profile:
 | Namespace (MCP server name) | Profile (role) | Typical use |
 |-----------------------------|----------------|-------------|
 | `gitea-author` | an author profile | implement issues, push branches, open PRs, comment |
-| `gitea-reviewer` | a reviewer profile | review, approve/request changes, merge |
+| `gitea-reviewer` | a reviewer profile | review, approve/request changes |
+| `gitea-merger` | a merger profile | merge PRs after approval and verification |
 | `gitea-reconciler` | a reconciler profile | close already-landed open PRs after ancestry proof (#304 profile; #310 close tool) |
 
+Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
+
 Properties:
 
 - **One process, one credential.** Each namespace authenticates as exactly
@@ -106,6 +109,14 @@ syntax to the client):
         "GITEA_MCP_CONFIG": "<path-to-profiles.json>",
         "GITEA_MCP_PROFILE": "<reviewer-profile-name>"
       }
+    },
+    "gitea-merger": {
+      "command": "<path-to>/venv/bin/python3",
+      "args": ["<path-to>/mcp_server.py"],
+      "env": {
+        "GITEA_MCP_CONFIG": "<path-to-profiles.json>",
+        "GITEA_MCP_PROFILE": "<merger-profile-name>"
+      }
     }
   }
 }

docs/gitea-execution-profiles.md

@@ -238,11 +238,43 @@ narrow operation set:
 - `gitea.issue.comment`
 - `gitea.issue.close`
 - `gitea.pr.close`
+- `gitea.branch.delete` (merged-branch cleanup only — see below)
 
 Forbidden on reconciler profiles: `gitea.pr.approve`, `gitea.pr.merge`,
 `gitea.pr.review`, `gitea.pr.create`, `gitea.branch.push`, and
 `gitea.repo.commit`.
 
+### Merged-branch cleanup ownership (`gitea.branch.delete`)
+
+The reconciler is the repository-supported owner of merged-PR source-branch
+cleanup: `task_capability_map` maps `cleanup_merged_pr_branch` (and
+`reconciliation_cleanup`) to role `reconciler` with permission
+`gitea.branch.delete`. Post-merge branch lifecycle is reconciliation work —
+it happens after the author, reviewer, and merger roles have completed, and
+it must not be reachable from those roles.
+
+Least-privilege constraints:
+
+- `gitea.branch.delete` is granted **only** to reconciler profiles. Author,
+  reviewer, and merger profiles must never hold it; `gitea_delete_branch`
+  and `gitea_cleanup_merged_pr_branch` fail closed on any profile without
+  the permission.
+- Even with the permission, reconciler deletion is only supported through the
+  guarded `gitea_cleanup_merged_pr_branch` path (#514 / #687): the PR must be
+  merged, the head an ancestor of the target, the branch not protected
+  (`master`/`main`/`dev`), the branch not a preservation/evidence ref (e.g.
+  `chore/issue-681-preserve-review-session-wip`), no open PR may still use the
+  head, and an explicit `CLEANUP MERGED PR <n> BRANCH <branch>` confirmation is
+  required. Raw `gitea_delete_branch` is **denied** to reconciler even when
+  `gitea.branch.delete` is present.
+- Raw `git branch -d` / `git push --delete` cleanup remains blocked by
+  `branch_cleanup_guard` and the final-report validator regardless of
+  profile permissions.
+- `gitea.branch.delete` has no short alias in `GITEA_OPERATION_ALIASES`;
+  write it fully qualified in `allowed_operations`. Migration must emit
+  canonical names such as `gitea.pr.close` (never bare `pr.close` /
+  `issue.close`, which the production normalizer rejects or drops).
+
 Launch a static `gitea-reconciler` MCP namespace with
 `GITEA_MCP_PROFILE=prgs-reconciler`. Profile shape is validated by
 `reconciler_profile.assess_reconciler_profile` (#304). Use the
@@ -251,6 +283,159 @@ Launch a static `gitea-reconciler` MCP namespace with
 fresh target-branch fetch, recorded target SHA, and ancestor proof. PRs whose
 heads are not already landed cannot be closed through this path.
 
+### Operational runbook: grant reconciler `gitea.branch.delete` (#687)
+
+Merging a code PR that updates `migrate_profiles.py` / `reconciler_profile.py`
+**does not** change the live operator profile on disk. Apply the profile
+change deliberately, then reconnect the client-managed namespace.
+
+1. **Approved migration / profile-update command** (from the repo root, using
+   the project venv if present):
+
+   ```bash
+   # Dry-run first (default): validates v2 output, writes nothing
+   python3 migrate_profiles.py -i ~/.config/gitea-tools/profiles.json
+
+   # Apply: creates backup then writes migrated v2 config
+   python3 migrate_profiles.py -i ~/.config/gitea-tools/profiles.json -w
+   # Optional explicit paths:
+   # python3 migrate_profiles.py -i ~/.config/gitea-tools/profiles.json \
+   #   -o ~/.config/gitea-tools/profiles.json \
+   #   --backup ~/.config/gitea-tools/profiles.json.bak -w
+   ```
+
+   If the live file is already v2, edit the reconciler identity’s
+   `allowed_operations` / `forbidden_operations` under
+   `environments.<env>.services.gitea.identities.reconciler` (or the
+   `prgs-reconciler` alias target) so allowed includes the canonical set
+   below — then re-validate with a load of the config (see step 3).
+
+2. **Inspect the generated (or edited) profile** — confirm the reconciler
+   identity, for example:
+
+   ```bash
+   python3 - <<'PY'
+   import json
+   from pathlib import Path
+   cfg = json.loads(Path.home().joinpath(".config/gitea-tools/profiles.json").read_text())
+   # v2 environments shape:
+   ident = cfg["environments"]["prgs"]["services"]["gitea"]["identities"]["reconciler"]
+   print("role:", ident.get("role"))
+   print("allowed:", ident.get("allowed_operations"))
+   print("forbidden:", ident.get("forbidden_operations"))
+   PY
+   ```
+
+3. **Validate canonical operation names and least privilege**
+
+   Expected canonical **allowed** (defaults after migration):
+
+   - `gitea.read`
+   - `gitea.pr.close` (required)
+   - `gitea.pr.comment`
+   - `gitea.issue.comment`
+   - `gitea.issue.close`
+   - `gitea.branch.delete` (recommended; cleanup only)
+
+   Expected **forbidden** includes at least: `gitea.pr.approve`,
+   `gitea.pr.merge`, `gitea.pr.review`, `gitea.pr.create`,
+   `gitea.branch.push`, `gitea.repo.commit`.
+
+   No shorthand (`pr.close`, `issue.close`, `pr.comment`) may remain.
+   Validate with the production loader:
+
+   ```bash
+   python3 - <<'PY'
+   import gitea_config, reconciler_profile
+   from pathlib import Path
+   path = str(Path.home() / ".config/gitea-tools/profiles.json")
+   gitea_config.load_config(path)  # fails closed on invalid config
+   # Or assess the reconciler lists directly after extracting them:
+   # print(reconciler_profile.assess_reconciler_profile(allowed, forbidden))
+   PY
+   ```
+
+4. **Merging PR #688 (or any code PR) does not update the live profile.**
+   Code changes only the migration helper, schema, docs, and tests. The
+   operator must still run `migrate_profiles.py -w` or an equivalent
+   authorized edit of `~/.config/gitea-tools/profiles.json`.
+
+5. **Supported apply method:** `python3 migrate_profiles.py … -w` (backup
+   created automatically) **or** operator-authorized edit of the live
+   profiles file after backup. Unsupported: silent mtime tricks, manual
+   process kill to “reload”, or undocumented env overrides.
+
+6. **Backup and validation:** `-w` copies the input to
+   `<input_path>.bak` (or `--backup PATH`) before writing. Re-run
+   `load_config` / `assess_reconciler_profile` after write. Keep the
+   `.bak` until live whoami/capability checks pass.
+
+7. **Client-managed namespace reconnect/reload:** reconnect or reload the
+   IDE MCP client so `gitea-reconciler` restarts from current `master` and
+   the updated `GITEA_MCP_PROFILE=prgs-reconciler` config. Do not hand-launch
+   `mcp_server.py` / `gitea_mcp_server.py` with ad hoc `GITEA_*` env
+   (see #686 / #630).
+
+8. **Live reverification** (through the client-managed `gitea-reconciler`
+   namespace only):
+
+   - `gitea_whoami` → identity + profile `prgs-reconciler`
+   - `gitea_assess_master_parity` → `stale=false`, `restart_required=false`
+   - `gitea_resolve_task_capability(task="cleanup_merged_pr_branch")` →
+     `allowed_in_current_session=true` only when permission and role match
+   - `gitea_resolve_task_capability(task="delete_branch")` →
+     **not** allowed for reconciler (role denial must be enforced)
+
+9. **Guarded cleanup usage** (example for a merged PR whose source branch
+   remains on the remote):
+
+   ```text
+   gitea_cleanup_merged_pr_branch(
+     pr_number=<N>,
+     branch=<exact PR head branch>,
+     confirmation="CLEANUP MERGED PR <N> BRANCH <exact PR head branch>",
+     remote="prgs",
+     org="Scaled-Tech-Consulting",
+     repo="Gitea-Tools",
+     worktree_path="<path under branches/>",
+   )
+   ```
+
+   The tool refuses unmerged PRs, protected branches, preservation/evidence
+   branches, open-PR heads, mismatched branch names, and wrong confirmation.
+
+10. **Prohibitions**
+
+    - No raw `git push --delete`, `git branch -d` / `-D`, or delete refspecs
+    - No arbitrary `gitea_delete_branch` from reconciler
+    - No unsupported profile switching mid-run without full re-preflight
+    - No ad hoc hand-edits of live profiles **unless** operator-authorized,
+      backed up, and revalidated as above
+
+Canonical migrated reconciler example:
+
+```json
+{
+  "role": "reconciler",
+  "allowed_operations": [
+    "gitea.read",
+    "gitea.pr.close",
+    "gitea.pr.comment",
+    "gitea.issue.comment",
+    "gitea.issue.close",
+    "gitea.branch.delete"
+  ],
+  "forbidden_operations": [
+    "gitea.pr.approve",
+    "gitea.pr.merge",
+    "gitea.pr.review",
+    "gitea.pr.create",
+    "gitea.branch.push",
+    "gitea.repo.commit"
+  ]
+}
+```
+
 ## Identity and fail-closed rules
 
 Before **any** mutating action, a workflow must know both:
@@ -311,7 +496,8 @@ To make Gitea MCP profile activation and runtime identity state explicit, the fo
 ### 2. Dual MCP Namespaces Recommendation
 For security-sensitive or high-risk tasks, the preferred safety model uses separate, isolated MCP server instances (namespaces/sessions) launched with static profiles:
 - `gitea-author`: Exposes tools configured with author permissions; cannot perform approvals or merges.
-- `gitea-reviewer`: Exposes tools configured with reviewer permissions; used for PR reviews and merges.
+- `gitea-reviewer`: Exposes tools configured with reviewer permissions; used for PR reviews. Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
+- `gitea-merger`: Exposes tools configured with merger permissions; used for PR merges.
 This layout maintains physical separation of credentials and prevents privilege escalation within a single session.
 This is the model accepted in #139; deployment details, rationale, and client
 setup live in

docs/label-taxonomy.md

@@ -1,34 +1,165 @@
 # Label Taxonomy
 
-This document catalogs the issue labels used for MCP workflows, including Jenkins and GlitchTip (observability).
+This document defines the canonical issue labels used by MCP workflows.
 
-> **Approval Required:** Do not create or apply new labels in `manage_labels.py` without explicit owner approval of this document.
+Every issue should carry:
 
-## Existing Labels
+- one `type:*` label
+- one `status:*` label
 
-* **`jenkins`**
-  * Description: Jenkins integration
-  * Color: `d93f0b`
-  * Use: Used to mark issues, PRs, or tasks that involve the `jenkins-mcp` boundaries, CI/CD designs, or build failures.
+Discussion-only issues must carry `type:discussion`.
 
-* **`glitchtip`**
-  * Description: GlitchTip integration
-  * Color: `b60205`
-  * Use: Used to mark issues related to the `glitchtip-mcp` boundary and observability integration.
+## Issue Type Labels
 
-## Proposed / Missing Labels
+| Label | Use |
+| --- | --- |
+| `type:bug` | Bug or defect |
+| `type:feature` | Feature or enhancement |
+| `type:process` | Process or policy work |
+| `type:workflow` | Workflow automation or guidance |
+| `type:guardrail` | Safety gate or guardrail |
+| `type:docs` | Documentation work |
+| `type:test` | Tests or test infrastructure |
+| `type:discussion` | Discussion-only issue |
+| `type:umbrella` | Umbrella or tracker issue |
+| `type:cleanup` | Cleanup or hygiene work |
 
-* **`observability`**
-  * Proposed Description: Observability, metrics, and monitoring tasks
-  * Proposed Color: `5319e7`
-  * Use: Broader than GlitchTip alone; covers logging, metrics, traces, and general observability pipeline improvements.
+## Workflow Status Labels
 
-* **`source:glitchtip`**
-  * Proposed Description: Issue filed automatically by GlitchTip orchestration
-  * Proposed Color: `b60205`
-  * Use: Applied automatically by the orchestrator when a GlitchTip error event is converted into a Gitea issue.
+Only one `status:*` label should be active on an issue at a time. When an issue
+moves forward, tooling must remove the old `status:*` label and apply the new
+one.
 
-* **`status:triage`**
-  * Proposed Description: Issue needs human or orchestrator triage
-  * Proposed Color: `fbca04`
-  * Use: Used for incoming issues (especially automated ones like `source:glitchtip`) that have not yet been evaluated for priority or resolution.
+| Label | Use |
+| --- | --- |
+| `status:triage` | Issue needs triage |
+| `status:ready` | Issue is ready for work |
+| `status:claimed` | Issue is claimed |
+| `status:in-progress` | Issue is being worked on |
+| `status:blocked` | Issue is blocked |
+| `status:needs-review` | Issue work needs review |
+| `status:pr-open` | A linked PR is open |
+| `status:changes-requested` | Reviewer requested changes on the linked PR |
+| `status:approved` | Linked PR is approved |
+| `status:merged` | Linked PR is merged |
+| `status:reconcile` | Issue needs reconciliation |
+| `status:done` | Issue workflow is complete |
+| `status:duplicate` | Issue is a duplicate |
+| `status:wontfix` | Issue will not be fixed |
+
+## Role Ownership Labels (#603)
+
+A single `role:*` label shows which workflow role currently owns the item. It is
+advisory visibility only — the control-plane lease (#601) is the source of truth
+for mutation authority. Only one `role:*` label is active at a time; tooling
+replaces it on handoff via `transition_role_labels`.
+
+| Label | Use |
+| --- | --- |
+| `role:author` | Author currently owns the item |
+| `role:reviewer` | Reviewer currently owns the item |
+| `role:merger` | Merger currently owns the item |
+
+## Hazard Labels (#603)
+
+Hazard labels are orthogonal warning flags. Unlike `status:*` and `role:*`,
+**more than one hazard may be active at once**, and a hazard never substitutes
+for a live lease / PR-state check. Add/remove with `add_hazard_label` /
+`clear_hazard_label`.
+
+| Label | Use |
+| --- | --- |
+| `hazard:stale-lease` | A stale or expired lease references this item |
+| `hazard:workflow-contaminated` | Session/workflow state is contaminated; do not mutate |
+| `hazard:conflicted` | Linked PR has merge conflicts |
+| `hazard:root-mutation` | Work was mutated in the project root checkout |
+| `hazard:manual-state` | Session or lease state was edited manually |
+| `hazard:terminal-blocker` | A terminal review/merge lock blocks progress (#332/#602) |
+
+Any item that carries `status:blocked` or any `hazard:*` flag must also have a
+blocking-reason / next-action comment (`requires_blocking_reason`).
+
+## `state:*` → canonical mapping (#603 migration)
+
+Issue #603 proposed a parallel `state:*` vocabulary. To avoid a conflicting
+second lifecycle prefix, those requested states are folded into the existing
+canonical labels rather than introduced as `state:*`. `state:*` is **not** a
+supported prefix; use the canonical label on the right.
+
+| Requested `state:*` | Canonical label |
+| --- | --- |
+| `state:needs-triage` | `status:triage` |
+| `state:claimed` | `status:claimed` |
+| `state:authoring` | `status:in-progress` |
+| `state:needs-review` | `status:needs-review` |
+| `state:reviewing` | `status:needs-review` |
+| `state:changes-requested` | `status:changes-requested` |
+| `state:approved` | `status:approved` |
+| `state:merge-ready` | `status:approved` |
+| `state:merged` | `status:merged` |
+| `state:blocked` | `status:blocked` |
+| `state:terminal-blocker` | `hazard:terminal-blocker` |
+| `state:abandoned` | `status:wontfix` |
+
+The transition helpers accept these names as synonyms (e.g.
+`canonical_status_label("authoring")` → `status:in-progress`), so callers may use
+the #603 wording while a single canonical status stays active.
+
+## Allocator Cross-Check (#603)
+
+Labels are advisory queue hints. The work allocator (#600/#613) uses labels as
+one signal but **cross-checks live leases and PR state** and never trusts labels
+alone. Discussion issues (`type:discussion`) are excluded from implementation
+queues (`is_implementation_candidate`) unless a controller explicitly selects
+them.
+
+## Transition Rules
+
+Suggested lifecycle:
+
+1. New issue created: `status:triage` or `status:ready`
+2. Issue selected by an author: `status:claimed`
+3. Author starts work: `status:in-progress`
+4. Work is blocked: `status:blocked`
+5. PR opened: `status:pr-open`
+6. PR approved: `status:approved`
+7. PR merged but issue still needs closure/reconciliation: `status:reconcile`
+8. Issue fully complete: `status:done`
+9. Duplicate issue: `status:duplicate`
+10. Won't-fix issue: `status:wontfix`
+
+The helper module `issue_workflow_labels.py` is the source of truth for the
+canonical label specs and status transition replacement behavior.
+
+## Discussion Issues
+
+Discussion issues must be labeled `type:discussion`.
+
+A discussion issue should not be treated as implementation-ready unless it also
+has a clear implementation status and next action.
+
+If a discussion produces implementation work, either:
+
+1. convert the discussion issue into an implementation issue by changing labels
+   and adding acceptance criteria, or
+2. create child implementation issues and leave the discussion issue as
+   `type:discussion`.
+
+## Tooling
+
+- `manage_labels.py --create-labels` creates the canonical `type:*` and
+  `status:*` labels.
+- `gitea_create_issue` recommends `type:*` and `status:*` labels when missing
+  and can apply supplied label names.
+- `gitea_mark_issue(..., action="start")` replaces old `status:*` labels with
+  `status:in-progress`.
+- `gitea_create_pr` fails closed before PR creation if `status:pr-open` cannot
+  be applied to the locked issue, then applies it after the PR is created.
+- `gitea_set_issue_labels` accepts an explicit `worktree_path` so author
+  sessions can satisfy the branches-only mutation guard while changing labels.
+
+## Existing Non-Workflow Labels
+
+Existing non-workflow labels such as `mcp`, `workflow`, `labels`, `tracker`,
+`jenkins`, `glitchtip`, `documentation`, and `testing` remain valid topical
+labels. They do not replace the required `type:*` and `status:*` labels.

docs/llm-workflow-runbooks.md

@@ -7,6 +7,11 @@ package of the MCP Control Plane: creating issues, implementing them, opening
 and reviewing pull requests, merging, and closing out — safely and
 reproducibly.
 
+Canonical state comments for durable issue/PR/discussion continuation are
+documented in [`canonical-state-comments.md`](canonical-state-comments.md).
+Use them when a workflow-changing comment needs to leave the next actor, next
+action, and paste-ready prompt in Gitea.
+
 > For the **project-agnostic** version of these operating rules (issue-first,
 > isolated worktrees, no self-review/merge, profile safety, cleanup, fail-closed)
 > that can be copied into any repository, see the reusable skill
@@ -28,6 +33,8 @@ audit logging). See [Related documents](#related-documents).
 > to discover the available project workflows and `mcp_get_skill_guide(<name>)`
 > for step-by-step instructions. This replaces long pasted operator prompts for
 > the standard rules; operator prompts still control task-specific scope.
+>
+> **BLOCKED + DIAGNOSE (default for any missing required step):** If a required workflow skill, guide, tool, capability, preflight, terminal, worktree binding, profile, or instruction is unavailable or fails, STOP. State BLOCKED. Use the canonical blocker report template (see skills/llm-project-workflow/templates/blocked-diagnose-report.md and the llm-project-workflow/SKILL.md universal rules). Only non-mutating recovery. Report fully. No unsafe fallbacks (temp scripts, direct API, MCP internals, direct imports, in-memory restoration, manual bypasses) unless controller authorizes in the handoff for this case. Missing required steps must fail closed *before* any git or Gitea mutation. Controller prompts and all workflows must reinforce: BLOCKED + DIAGNOSE, then stop.
 > See issue #129 for the skill registry design.
 
 Jenkins and GlitchTip workflows use separate MCP servers, not this Gitea MCP
@@ -188,7 +195,7 @@ To avoid the bottleneck of relaunching/restarting the MCP server to switch betwe
   `gitea_reconcile_already_landed_pr` after ancestry proof — not for normal
   review or author workflows.
 
-* **Fallback:** If the dual-profile MCP launcher pattern is not supported or configured in the client, the LLM must relaunch or restart the client/MCP with the correct profile environment variable before claiming or working on any tasks.
+* **Fallback (operator-owned):** If the dual-profile MCP launcher pattern is not supported or configured in the client, **do not** have the LLM relaunch or restart the client/MCP. The LLM **stops** role-switching work, reports that the correct static namespace is missing, and waits for an **operator** to configure dual namespaces or reload the client with the correct `GITEA_MCP_PROFILE` for that role. Process restart/relaunch is operator-owned under the [stable control runtime ADR](architecture/mcp-stable-control-runtime-policy-adr.md) (#615).
 
 ## Setup runbook — interactive menu
 
@@ -408,6 +415,31 @@ The guard blocks when the **control checkout** (repository root) is:
 `branches/...` directories are disposable role worktrees; the root checkout is
 the stable orchestration surface only.
 
+## Canonical workflow skill names (#551)
+
+Controller prompts and sessions must load the **same** workflow skill wall
+regardless of runtime (Claude, Codex, Gemini):
+
+| Name | Role |
+|------|------|
+| `gitea-workflow` | Primary controller / Codex skill name |
+| `llm-project-workflow` | Portable in-repo package |
+| `git-pr-workflows` | Legacy alias |
+
+- Inventory: `mcp_list_project_skills` lists all three.
+- Preflight: `mcp_check_workflow_skill_preflight` before mutations.
+- Codex install: `scripts/install-codex-workflow-skill.sh`
+- Full doc: [`docs/workflow-skill-mount.md`](workflow-skill-mount.md)
+
+If the skill is missing, stop with BLOCKED + DIAGNOSE — do not mutate.
+
+## No direct-import mutation path (#558)
+
+Never `import gitea_mcp_server` or call `gitea_auth.get_auth_header` /
+keychain fill from a raw shell to bypass MCP preflight.
+
+Use the official MCP daemon only. See [`docs/mcp-daemon-import-guard.md`](mcp-daemon-import-guard.md).
+
 ## Bootstrap Review Path for self-hosted MCP fixes (#557)
 
 When a PR fixes Gitea-Tools MCP workflow code that the **live daemon** still
@@ -629,24 +661,44 @@ session, clearing hung background terminals, switching to MCP-native commit, and
 the agent temp artifact cleanup checklist. Do **not** retry shell encoding in a
 loop and do **not** substitute WebFetch/Playwright/manual base64.
 
+### Terminal launcher diagnostics (#556)
+
+When git/pytest finalization fails with opaque spawn errors (e.g. `os error 2`),
+do **not** improvise shell wrappers or fall back to direct API / temp scripts.
+
+1. Call `gitea_diagnose_terminal` (optional `cwd`, optional `command`) — returns
+   categorized failure: missing cwd, cwd not a directory, missing executable,
+   missing runtime wrapper, missing shell, probe timeout, or session launcher
+   failure.
+2. Call `gitea_get_shell_health` for the shell circuit-breaker state.
+3. Mutation-capable `gitea_resolve_task_capability` probes the terminal launcher
+   before allowing git/pytest-oriented work; on failure it returns
+   `BLOCKED + DIAGNOSE` with `terminal_launcher_unhealthy` and diagnostics.
+4. Emit the canonical `blocked-diagnose-report.md` template and stop.
+
 ### Create an issue / child issues
 
 - **Profile:** issue-manager or author (any profile allowed to create issues).
 - **Steps:** create the parent/roadmap issue; create child issues; apply the
   minimal label set; link children to the parent.
+- **Labels:** new issues should carry one `type:*` label and one `status:*`
+  label. Discussion-only issues must carry `type:discussion`. See
+  [`label-taxonomy.md`](label-taxonomy.md).
 - **Prompt:** `Using the issue-manager profile, create issue "<title>" with body
   <body>, then create child issues for <list> and link them to the parent.`
 
 ### Implement an issue and open a PR
 
 - **Profile:** author.
-- **Steps:** claim the issue (`status:in-progress`); create an isolated branch
-  worktree from latest `master` under `branches/` (`feat/issue-<n>-...` /
+- **Steps:** claim the issue (`status:in-progress`, replacing any old
+  `status:*` label); create an isolated branch worktree from latest `master`
+  under `branches/` (`feat/issue-<n>-...` /
   `fix/...` / `docs/...`); `cd` into that worktree; implement narrowly; add or
   update tests if behavior changes; run the full suite; commit with an
-  issue-linked message; open a PR to `master`. **Do not** review or merge your
-  own PR. Include an `LLM Handoff Metadata` block (with `LLM-Agent-SHA`) in
-  the PR body — see [`llm-agent-sha.md`](llm-agent-sha.md).
+  issue-linked message; open a PR to `master`; move the issue to
+  `status:pr-open`. **Do not** review or merge your own PR. Include an
+  `LLM Handoff Metadata` block (with `LLM-Agent-SHA`) in the PR body — see
+  [`llm-agent-sha.md`](llm-agent-sha.md).
 - **Prompt:** `Use an author profile to implement issue #N and open a PR to
   master. Do not self-review or self-merge.`
 
@@ -687,10 +739,10 @@ loop and do **not** substitute WebFetch/Playwright/manual base64.
 ### Merge a PR
 
 - **Profile:** merger (allowed to merge; must **not** be the PR author).
-- **Steps:** confirm eligibility; require explicit confirmation
-  (`MERGE PR <n>`); optionally pin head SHA / changed-file set; merge only when
-  Gitea reports the PR mergeable (branch-protection checks satisfied). No force,
-  no ignore-checks. Verify that remote master contains the merge commit or the expected squashed changes (do not assume a "closed" PR succeeded without verifying the actual landed changes).
+- **Steps:**
+  - Merger workflow starts only after formal approval at current head. Reviewer workflow ends with review decision and separate merger handoff.
+  - Confirm eligibility; require explicit confirmation (`MERGE PR <n>`); optionally pin head SHA / changed-file set; merge only when Gitea reports the PR mergeable (branch-protection checks satisfied). No force, no ignore-checks. Verify that remote master contains the merge commit or the expected squashed changes (do not assume a "closed" PR succeeded without verifying the actual landed changes).
+  - Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
 - **Prompt:** `Use any eligible merger profile to merge PR #N if checks pass and
   it is mergeable. Confirm with "MERGE PR N". Do not force-merge.`
 
@@ -738,6 +790,53 @@ merger handoff.
 - **Prompt (normal):** `After verifying master contains the merge of PR #N using post-merge file-presence verification, close issue #M and delete the merged branch. Include verification details in the report.`
 - **Prompt (reconcile):** `Reconcile closed-not-merged PR #N by verifying if its content landed on master.`
 
+### Post-merge merged cleanup ownership (#523)
+
+Post-merge **local worktree / remote branch cleanup** is **reconciler** work, not
+author work. Do not switch from `prgs-reconciler` to `prgs-author` only to run
+`gitea_reconcile_merged_cleanups`.
+
+- **Profile:** `prgs-reconciler` (task `reconcile_merged_cleanups` /
+  `reconciliation_cleanup`).
+- **Namespace:** reconciler MCP server; stable control checkout is allowed for
+  this role (branches-only author guard does not apply).
+- **Steps:**
+  1. `gitea_whoami` + `gitea_resolve_task_capability(task="reconcile_merged_cleanups")`.
+  2. Dry-run first: `gitea_reconcile_merged_cleanups(dry_run=True)`.
+  3. Execute only after audit/authorization gates when remote branch delete or
+     worktree removal is required (`dry_run=False`, `execute_confirmed=True`,
+     and `gitea.branch.delete` when deleting remotes).
+- **Fail closed:** unmerged/open heads, mismatched worktrees, and non-merged
+  closed PRs must not be cleaned.
+- **Reports:** label cleanup actions as reconciler cleanup (not author mutation).
+- **Prompt:** `As prgs-reconciler, dry-run then execute gitea_reconcile_merged_cleanups for recently merged PRs without switching to prgs-author.`
+
+### Superseded PR / satisfied issue reconciliation (#525)
+
+Closing duplicate or superseded PRs after a canonical PR has merged is
+**reconciler** work. Do not switch to `prgs-author` only to close the duplicate
+PR, close the satisfied issue, or file a narrow follow-up discovered during
+reconciliation.
+
+- **Profile:** `prgs-reconciler`.
+- **Tasks:** `reconcile_close_superseded_pr`,
+  `reconcile_close_satisfied_issue`, and
+  `reconcile_create_followup_issue`.
+- **Tool:** `gitea_reconcile_superseded_by_merged_pr`.
+- **Required proof:**
+  1. Live target PR state is open, but it is not independently required and no
+     mergeable work remains.
+  2. Live superseding PR state is closed and merged.
+  3. Superseding PR head is an ancestor of freshly fetched `master`.
+  4. Merge commit SHA is recorded.
+  5. Canonical close comment cites the superseding PR and merge commit.
+  6. Linked issue closure is attempted only when the issue is open and
+     explicitly satisfied by the superseding PR.
+- **Fail closed:** missing ancestry proof, missing canonical comment,
+  mergeable target PR, same target/superseding PR, or missing
+  `gitea.pr.close` / `gitea.issue.close` capability.
+- **Prompt:** `As prgs-reconciler, reconcile PR #N as superseded by merged PR #M; post the canonical close comment, close the superseded PR, and close issue #K only if the merged PR fully satisfies it.`
+
 ### Stop on blocker
 
 - **Any profile.** If a required gate cannot be satisfied — identity
@@ -758,7 +857,7 @@ an author.
 |---|---|---|---|---|
 | Review PR (`review_pr`) | reviewer (e.g. `sysadmin` / `prgs-reviewer`) | read, gated review verdicts | commits, pushes, file edits, author comments, merge without eligibility | active profile is an author profile — stop immediately; do **not** switch to author-side fixes unless the operator explicitly re-tasks |
 | Address PR change requests (`address_pr_change_requests`) | author (e.g. `jcwalker3` / `prgs-author`) | commit/push fixes to the PR branch, PR comment summarizing fixes | review verdicts, approve, request-changes, merge | active profile lacks branch push |
-| Merge PR (`merge_pr`) | reviewer/merger | gated merge after eligibility + approval | merging own PR, merging without pinned head match | active profile is an author profile, or any merge gate fails |
+| Merge PR (`merge_pr`) | merger (e.g. `sysadmin` / `prgs-merger`) | gated merge after eligibility + approval | merging own PR, merging without pinned head match, reviewing PRs | active profile is an author or reviewer-only profile, or any merge gate fails |
 | Comment on issue discussion (`comment_issue`) | any profile with `gitea.issue.comment` | issue thread comments | review verdicts, closing via comment | permission missing (`gitea.pr.comment` does **not** imply it) |
 | Comment on PR (`comment_pr`) | any profile with `gitea.pr.comment` | PR thread comments | review verdicts | permission missing |
 | Author implementation (`create_branch`/`push_branch`/`create_pr`) | author | branch, commit, push, open PR | self-review, self-merge | profile lacks the author permissions |
@@ -808,6 +907,27 @@ Never imply full-suite success unless the full-suite command itself passed
 (`full_suite_passed: true`). A report that hides a failed or skipped check
 is worse than a failing report.
 
+## Canonical Thread Handoff (CTH)
+
+**CTH** = **Canonical Thread Handoff** — the authoritative workflow handoff
+comment in a Gitea issue or PR thread. See
+[`canonical-thread-handoff.md`](canonical-thread-handoff.md) for types,
+templates, and examples.
+
+Before acting in an issue or PR thread:
+
+1. **Find the latest CTH comment** before doing work.
+2. Treat the **latest valid CTH** as the current handoff state.
+3. **Post a new CTH** when you finish, block, skip, supersede, request
+   changes, approve, or hand off.
+4. Do **not** rely on stale non-CTH comments when a newer CTH exists.
+
+A CTH summarizes workflow state for the next session. Formal Gitea review
+verdicts remain authoritative for merge gates — a CTH is not merge approval
+by itself.
+
+Template: [`../skills/llm-project-workflow/templates/canonical-thread-handoff.md`](../skills/llm-project-workflow/templates/canonical-thread-handoff.md)
+
 ## Controller Handoff (required, every task)
 
 Every task — implementation, review, merge, triage, documentation,
@@ -840,6 +960,161 @@ touched release state names the exact tag/commit and why. Design debates
 belong in **discussion/RFC issues** (e.g. #100 `profiles.json v2`) — comment
 on the issue, create no branches/PRs, and end the comment with this handoff.
 
+## Two-comment workflow reporting (#507)
+
+After meaningful controller/workflow work, post **two separate Gitea comments**
+(not one combined blob):
+
+1. **`[CONTROLLER HANDOFF]`** — detailed operational continuation for the
+   next LLM/controller (proof-heavy; may be long).
+2. **`[THREAD STATE LEDGER]`** — short canonical truth readable in ~30 seconds.
+
+The ledger must answer: what is true now, what changed, what is blocked,
+who/what acts next — and must **separate**:
+
+- local verdict/state
+- server-side Gitea state
+- attempted-but-blocked mutations
+- completed mutations
+
+Use precise state phrases (`APPROVED review posted to Gitea`,
+`APPROVE verdict prepared locally`, `merge performed`, `merge not performed`,
+`no server-side state changed`, `lease attempt blocked`) instead of ambiguous
+standalone words (`approved`, `merged`, `ready`, `blocked`, `done`).
+
+The ledger must include a **blocker classification** from:
+`code blocker`, `test blocker`, `merge conflict`, `stale head`,
+`permission/capability blocker`, `environment/tooling blocker`,
+`process/rule blocker`, `queue/lease blocker`,
+`duplicate/canonicalization blocker`, `no blocker`.
+
+Templates: [`two-comment-workflow.md`](two-comment-workflow.md).
+Worked examples: [`examples/two-comment-workflow-examples.md`](examples/two-comment-workflow-examples.md).
+
+Validation: `thread_state_ledger_validator.py` checks tagged comments at post
+time (`gitea_create_issue_comment`) and tagged final reports via
+`assess_final_report_validator`. Legacy `## Controller Handoff` final reports
+remain valid during transition; the tagged pair is required for new workflow
+comments.
+
+Related (do not duplicate): #494/#495 lifecycle state, #501 mutation-ledger
+consistency, #505 CTH umbrella, #496 workflow comment gate when merged.
+
+## Canonical comment validation (#496)
+
+Workflow-changing issue/PR/review comments must carry durable next-action
+state. Casual discussion is still allowed.
+
+The MCP server runs `canonical_comment_validator.assess_canonical_comment`
+**before** posting through:
+
+- `gitea_create_issue_comment`
+- `gitea_submit_pr_review` / `gitea_dry_run_pr_review` (non-empty review bodies)
+- `gitea_reconcile_already_landed_pr` when `post_comment=True`
+- internal structured comment helpers (machine lease/heartbeat markers stay exempt)
+
+Detection examples:
+
+- **Allowed:** `Thanks, I will check this.`
+- **Rejected:** `Blocked, author should fix.` (workflow trigger without canonical fields)
+
+When validation fails, the tool returns `canonical_comment_validation` with
+`allowed: false`, `missing_fields`, `vague_fields`, `correction_message`, and
+`suggested_template`. **No Gitea API call is made.**
+
+Minimum workflow comment fields:
+
+```text
+STATE:
+WHO_IS_NEXT:
+NEXT_ACTION:
+NEXT_PROMPT:
+WHY:
+```
+
+`WHO_IS_NEXT` must be one of: `controller`, `author`, `reviewer`, `merger`,
+`reconciler`, `user`.
+
+Issue comments also require `RELATED_PRS`, `BLOCKERS`, and `VALIDATION` when
+they mention PR work. PR comments/reviews also require `ISSUE`, `HEAD_SHA`,
+`REVIEW_STATUS`, `MERGE_READY`, `BLOCKERS`, and `VALIDATION`.
+
+Special states:
+
+- `STATE: blocked` — `BLOCKERS` must name an explicit unblock condition.
+- `STATE: superseded` — requires `CANONICAL_ITEM` and `SUPERSEDED_ITEM`.
+- `STATE: ready-to-merge` — requires approval proof and head SHA in
+  `HEAD_SHA`, `REVIEW_STATUS`, `MERGE_READY`, or `VALIDATION`.
+
+Final reports must not claim a comment was posted when
+`canonical_comment_validation.allowed` is false (#496 AC14).
+
+## Stale #332 review-decision lock cleanup (#594)
+
+#332 hard-stops a reviewer session after a terminal live review mutation
+(`approve` / `request_changes`). After #559 those locks are **durable** under
+`~/.cache/gitea-tools/session-state/review_decision_lock-<profile>.json`, so they
+can outlive the PR they protected.
+
+### When cleanup is **allowed**
+
+Use `gitea_cleanup_stale_review_decision_lock` from a **reviewer** profile when:
+
+1. A durable review-decision lock has a terminal live mutation, **and**
+2. Live Gitea state shows that terminal PR is already **merged** or **closed**
+   (so no same-PR merge sequence remains), **and**
+3. The active profile identity matches the lock, **and**
+4. Authenticated identity can be verified.
+
+Workflow:
+
+```text
+# 1) Assess only (default)
+gitea_cleanup_stale_review_decision_lock(apply=false, remote=prgs, ...)
+
+# 2) Apply after is_moot=true and cleanup_allowed=true
+gitea_cleanup_stale_review_decision_lock(
+  apply=true,
+  expected_terminal_pr=<merged-or-closed-pr>,
+  remote=prgs,
+  ...
+)
+```
+
+Successful apply:
+
+* clears in-memory + durable decision lock for that profile
+* returns a structured `audit` payload
+* posts an audit comment on the mooted PR when `gitea.pr.comment` is allowed
+  (`post_audit_comment=true` default)
+
+Same-profile auto-expire: if `gitea_merge_pr` succeeds on a PR that this same
+profile just approved, the decision lock is cleared after merge. Cross-profile
+locks (reviewer vs merger) still require the cleanup tool.
+
+### When cleanup is **forbidden**
+
+Do **not** clear when:
+
+* the last terminal PR is still **open** (active #332 hard-stop — continue merge
+  for that approve, or stop after request_changes)
+* live PR state cannot be fetched (fail closed)
+* profile identity does not match the lock
+* identity cannot be verified
+* `expected_terminal_pr` does not match the last terminal PR
+
+**Never** use manual deletion of session-state files as the normal workflow.
+**Never** use cleanup to skip review of an open PR or to bypass eligibility /
+mark-ready / submit gates on active work.
+
+### Related tools
+
+| Tool | Purpose |
+|------|---------|
+| `gitea_cleanup_stale_review_decision_lock` | Moot decision-lock cleanup (#594) |
+| `gitea_authorize_review_correction` | Operator-approved correction after a **mistaken** live review on still-active work (#211) |
+| `gitea_cleanup_post_merge_moot_lease` | Moot **lease** cleanup after merge (#515) — different object |
+
 ## Fail-closed behavior
 
 Before any mutating action the workflow verifies identity, active profile,
@@ -925,10 +1200,13 @@ When a mutation blocks on workspace binding:
 
 1. Read the error — it names the **resolved workspace path**, **role
    namespace**, and **binding source** (tool arg, env var, or process root).
-2. Reconnect or relaunch the correct namespace MCP server from the intended
-   workspace (or set the role-specific env var before launch).
-3. Pass `worktree_path` on reviewer/merger mutation tools when the active
-   branches/ worktree differs from the MCP process root.
+2. **LLM-allowed:** pass `worktree_path` on reviewer/merger mutation tools when
+   the active `branches/` worktree differs from the MCP process root; **client
+   reconnect** after transport EOF only (no process kill).
+3. **Operator-owned:** if the wrong namespace process was launched, or a role-
+   specific `GITEA_*_WORKTREE` must be set at process start, an **operator**
+   reloads/relaunches the correct static namespace MCP. LLM sessions must not
+   kill or restart MCP processes (see [stable control runtime ADR](architecture/mcp-stable-control-runtime-policy-adr.md)).
 4. **Do not** clean, reset, or discard foreign role worktrees to unblock your
    own namespace — that destroys another agent's WIP.
 
@@ -938,9 +1216,10 @@ When posting a Canonical Thread Handoff after a binding blocker:
 
 - State which namespace was active (author / reviewer / merger / reconciler).
 - Quote the resolved workspace path and binding source from the error.
-- Name the safe reconnect action (relaunch MCP from `branches/...`, set
-  `GITEA_*_WORKTREE`, or pass `worktree_path`).
-- Explicitly note that foreign worktrees must not be cleaned to unblock.
+- Name the safe next action: pass `worktree_path` if that unblocks the tool, or
+  request an **operator** reload of the correct namespace MCP / env binding.
+- Explicitly note that foreign worktrees must not be cleaned to unblock, and
+  that the LLM must not self-restart the MCP process.
 
 ## Safety notes
 
@@ -951,6 +1230,8 @@ When posting a Canonical Thread Handoff after a binding blocker:
 
 ## Related documents
 
+- [`architecture/mcp-stable-control-runtime-policy-adr.md`](architecture/mcp-stable-control-runtime-policy-adr.md) — stable control runtime vs dev runtime; LLM must not kill/restart MCP; operator-owned reload and promotions; routine post-merge parity staleness (#615).
+- [`reviewer-handoff-consistency.md`](reviewer-handoff-consistency.md) — reject contradictory reviewer handoffs (#501).
 - [`issue-acceptance-gate.md`](issue-acceptance-gate.md) — controller issue-acceptance audit after PR merge (#500).
 - [`../skills/llm-project-workflow/SKILL.md`](../skills/llm-project-workflow/SKILL.md) — portable cross-project LLM workflow skill.
 - [`gitea-execution-profiles.md`](gitea-execution-profiles.md) — the profile model.

docs/mcp-daemon-import-guard.md

@@ -0,0 +1,92 @@
+# MCP daemon import and native-transport guard (#558 / #695)
+
+## Problem
+
+During deadlock debugging and the PR #694 incident (#695), agents imported
+`gitea_mcp_server` or ran credential helpers from a raw shell / offline helper,
+bypassing native MCP transport, preflight purity, and role gates. Contaminated
+formal reviews then looked identical to native approvals.
+
+## Rule
+
+Mutation auth, keychain fill, and controller quarantine require a **production
+native MCP transport runtime** established only by:
+
+1. the **resolved absolute path** of the canonical entrypoint
+   (`mcp_server.py` / `gitea_mcp_server.py` next to `mcp_daemon_guard.py`), and
+2. a live **transport bind** (`bind_native_mcp_transport(transport="stdio")`)
+   immediately before `mcp.run`.
+
+Basename-only trust (a renamed file called `mcp_server.py`), caller-controlled
+flags (there is **no** `allow_test_bootstrap`), environment variables, stack
+frame spoofing, or import-only launch are insufficient.
+
+| Context | Allowed |
+|---------|---------|
+| Official IDE-native MCP: resolved canonical entrypoint marks + binds stdio, holds process-local runtime token | yes |
+| pytest (hermetic unit tests) via `is_pytest_runtime()` | yes for unit gates |
+| `install_test_native_runtime()` under pytest (test-mode record) | unit-test transport gates only — **never** production Gitea mutations |
+| `allow_test_bootstrap=True` (removed; must not exist) | **no** |
+| Renamed runner basename `mcp_server.py` outside package root | **no** |
+| Import/launch of real entrypoint without transport bind | **no** |
+| `GITEA_MCP_SANCTIONED_DAEMON=1` alone (no process-local native runtime) | **no** (#695) |
+| `GITEA_ALLOW_DIRECT_MCP_IMPORT=1` in LLM sessions | **no** — never set; never authorizes mutations (#695 AC1 / PR #701) |
+| Override `GITEA_MCP_SESSION_STATE_DIR` mid-session | **no** — production bind pins state root; redirect cannot forge independent decision locks (#695 AC2 / PR #701) |
+| `GITEA_ALLOW_KEYCHAIN_CLI=1` in LLM sessions | **no** — human operator only |
+| bare `python -c 'import gitea_mcp_server; …'` or offline runners | **no** |
+| keychain fill outside native/pytest | **no** |
+
+Native runtime is **process-local**: a random token bound to the daemon PID and
+transport phase. It is never reconstructed from environment variables,
+session-state files, caller-controlled flags, or importing internals in a
+fresh Python process.
+
+## Contaminated review quarantine (#695 AC8)
+
+Controller/reconciler/merger profiles may call
+`gitea_quarantine_contaminated_review` with explicit confirmation:
+
+```text
+QUARANTINE CONTAMINATED REVIEW <review_id> PR <pr_number>
+```
+
+Quarantine records are durable under the MCP session-state root and are
+**honored** by:
+
+- `gitea_get_pr_review_feedback` (quarantined approvals do not authorize merge)
+- `gitea_check_pr_eligibility` action=`merge`
+- `gitea_merge_pr` (mutation)
+- merger lease adoption paths that read `approval_at_current_head`
+
+Forensic Gitea reviews and historical comments are **never deleted**.
+
+## STOP after native MCP failure (AC10)
+
+If the native MCP namespace dies (EOF, capability disconnect, session death):
+
+1. **STOP.** State BLOCKED + DIAGNOSE.
+2. Do **not** import `gitea_mcp_server` from a standalone process.
+3. Do **not** run `offline_mcp_helper.py`, `offline_mcp_runner.py`,
+   `run_quarantine.py`, or any offline mutation helper.
+4. Do **not** set direct-import, keychain-bypass, or raw-token environment
+   variables.
+5. Reconnect / restart the official MCP daemon; resume only via native tools.
+
+Any further native MCP failure is a hard stop. Do not construct another fallback.
+
+## Canonical approval claims (AC7)
+
+Comments that claim `approved` / `ready-to-merge` / `WHO_IS_NEXT: merger` /
+`MERGE_READY: true` must include:
+
+```text
+NATIVE_REVIEW_PROOF: transport=native_mcp; …
+```
+
+Claims that cite offline/import helpers are rejected even if a proof line is
+present.
+
+## Operator note
+
+LLM sessions must never set allow-direct-import, allow-keychain-cli, or raw
+token overrides. Those are human-only escape hatches outside agent workflows.

docs/mcp-menu.md

@@ -41,7 +41,7 @@ The script must be executable (`chmod +x mcp-menu.sh`). It uses bash with
 |--------|-------------|
 | Project status / root checkout health | Shows cwd, branch, `git status --short --branch`, HEAD SHA, `prgs/master` SHA, and warnings when the root checkout is dirty or off `master`. |
 | Author workflow prompts | Ready-to-copy prompts for issue work, conflict-fix sessions, and root checkout recovery. |
-| Reviewer workflow prompts | PR review prompt (review-only; no merge). |
+| Reviewer workflow prompts | Standard PR review prompt, and a skip-already-reviewed-stale-`REQUEST_CHANGES` prompt that hands off to the author without a duplicate terminal mutation (review-only; no merge). |
 | Merger workflow prompts | PR merge prompt (merge gates and explicit approval). |
 | Reconciler workflow prompts | Already-landed / closed PR reconciliation prompt. |
 | Onboarding new project | Checklist prompt for adding a repository to the MCP workflow. |

docs/mcp-namespace-eof-recovery.md

@@ -0,0 +1,118 @@
+# Recovering from `client is closing: EOF` on a Gitea MCP namespace (#543)
+
+## Symptom
+
+A tool call through a Gitea MCP namespace — `gitea-author`, `gitea-reviewer`,
+`gitea-merger`, or the shared `gitea-tools` namespace — fails immediately with:
+
+```
+client is closing: EOF
+```
+
+Every subsequent call to that same namespace returns the same error, including
+cheap read tools such as `gitea_whoami` and `gitea_list_profiles`. Other MCP
+servers registered with the same client (for example `context7`) keep working,
+so this is **not** a global MCP-client outage.
+
+## Why this is not a code defect
+
+This failure is a **transport-level** condition in the IDE / MCP client manager,
+not a missing or broken tool:
+
+- The tool can be present and registered in the Python `FastMCP` tool manager.
+- Direct Python inspection of the server confirms the tool exists.
+- Running the server manually and sending JSON-RPC over stdio works fine
+  (offline spawn) — that path does **not** prove the IDE namespace is healthy.
+
+The client manager entered a closed state after the backing subprocess for that
+namespace terminated (or was killed) behind its back. Once closed, the client
+does **not** re-spawn the child on the next tool call — it just replays
+`client is closing: EOF`. The OS process may even still be alive if a parent
+language-server process is holding the stdio pipes open.
+
+This is the canonical "registered in FastMCP ≠ callable through the namespace"
+false-ready state. It is distinct from the **stale-runtime** family in #531 /
+#544, where the process is reachable but running behind `master`; that case is
+detected by the `ps`-based `_check_mcp_runtimes_diagnostics` in
+`gitea_mcp_server.py`. The EOF case is a dead/closed transport, not a stale one,
+so the `ps` check alone will not surface it.
+
+## Recovery path (canonical — client reconnect only)
+
+Do the steps in order. Stop as soon as a live **client-namespace** call succeeds.
+
+1. **Confirm the blast radius.** Call a cheap read tool on the failing namespace
+   (`gitea_whoami` or `gitea_list_profiles`). Then call the same tool on a
+   different MCP server (e.g. `context7`).
+   - Only the Gitea namespace fails → single-namespace transport close. Continue.
+   - Every server fails → restart the whole MCP client, not just one namespace.
+
+2. **Reconnect the namespace through the client, not the shell.** Use the IDE /
+   client MCP-reconnect action for that server entry (in Claude Code:
+   `/mcp` → reconnect the affected `gitea-*` server). Reconnecting forces the
+   client to spawn a fresh subprocess and re-open the pipe. This clears the
+   closed-client state that a bare `kill`/respawn from a terminal does **not**.
+
+3. **Do not "fix" it by importing the server or poking the process.** Reaching
+   for `python -c 'import gitea_mcp_server ...'`, raw JSON-RPC from a shell,
+   killing PIDs to force a respawn, or touching MCP config mtimes does **not**
+   restore the *client's* view of the namespace and violates the daemon-import
+   guard (#558, `docs/mcp-daemon-import-guard.md`). The only sanctioned repair
+   is a **client reconnect / relaunch**.
+
+4. **Verify through the same path the workflow will use.** After reconnect, call
+   the specific tool the blocked workflow needs — not just any tool — through
+   the target namespace. For a merge that means calling the merger-authorized
+   adoption/merge tool through `gitea-merger`. A green `gitea_whoami` on one
+   namespace does **not** prove another namespace or another tool is callable.
+   Record success with:
+
+   ```text
+   gitea_assess_mcp_namespace_health(..., probe_source="client_namespace")
+   ```
+
+5. **If reconnect does not clear it,** relaunch the client entirely, then repeat
+   step 4. If EOF persists after a full relaunch, the backing subprocess is
+   failing to start — inspect its stderr / launch config (command path, venv,
+   `*_MCP_CONFIG`, `*_MCP_PROFILE` env) rather than retrying the call. Still
+   do not use PID kill or config-touch as the primary recovery.
+
+## Diagnostics to capture when reporting EOF
+
+Include all of these so the failure is actionable and reproducible:
+
+- **Namespace name** that returned EOF (`gitea-author` / `gitea-reviewer` /
+  `gitea-merger` / `gitea-tools`).
+- **Tool** that was called and the **exact** error string.
+- **PID** of the backing process (if any) and whether it was still alive
+  (informational only — not a recovery action).
+- **Profile / env** for that namespace (execution profile, `*_MCP_PROFILE`,
+  worktree binding such as `GITEA_AUTHOR_WORKTREE`).
+- **Config path** the client launched the server from.
+- Result of the **cross-server control** call (did `context7` succeed?).
+
+## Offline spawn probe (non-authoritative)
+
+`test_mcp_conn.py` performs a full JSON-RPC handshake against a **fresh
+subprocess** (`initialize` → `initialized` → `tools/list` → `tools/call`) and
+classifies with `probe_source=offline_spawn`. That is useful for offline
+launch/registration debugging. It is **not** proof the IDE-managed namespace is
+healthy. See `docs/mcp-namespace-health.md`.
+
+## Do-not list during EOF recovery
+
+- Do **not** retry a blocked merge/adoption until the required tool is confirmed
+  callable through the merger-authorized **client** namespace (see #543).
+- Do **not** clean, reset, or rebind a **foreign** worktree to work around the
+  error.
+- Do **not** bypass the namespace with direct imports, raw API/curl, or
+  in-memory state restoration.
+- Do **not** kill MCP PIDs or touch config mtimes as a substitute for client
+  reconnect.
+
+## Related
+
+- #531 / #544 — stale-runtime detection (`ps`-based); sibling failure mode.
+- #558 / `docs/mcp-daemon-import-guard.md` — why shell imports are not a repair.
+- `docs/mcp-client-registration.md` — per-server registration contract.
+- `docs/mcp-namespace-health.md` — probe sources and mutation enforcement.

docs/mcp-namespace-health.md

@@ -0,0 +1,88 @@
+# MCP namespace health diagnostics (#543)
+
+Gitea MCP tools can be registered in the Python FastMCP server while the IDE's
+live MCP namespace is still unusable. The failure usually appears as
+`client is closing: EOF`, `transport closed`, or an empty response when calling
+a tool such as `gitea_whoami`.
+
+Do not treat static tool registration as proof that review or merge workflows
+can proceed. A reviewer or merger flow must have **client-namespace** evidence
+that the required tool is callable through the configured IDE MCP namespace.
+
+## Probe sources (do not confuse them)
+
+| Source | How obtained | Proves IDE namespace? |
+| --- | --- | --- |
+| `client_namespace` | Tool call through the IDE-managed MCP client | **Yes** |
+| `offline_spawn` | `test_mcp_conn.py` subprocess JSON-RPC handshake | **No** (offline only) |
+
+`gitea_assess_mcp_namespace_health` accepts `probe_source` and only sets
+`ide_namespace_proven=true` for `client_namespace` success. Offline spawn
+success never clears review/merge mutation gates.
+
+## Client-namespace health check (canonical)
+
+1. Through the IDE client, call a cheap tool on the target namespace
+   (`gitea_whoami` or `gitea_list_profiles`).
+2. Feed the live result into:
+
+```text
+gitea_assess_mcp_namespace_health(
+    namespace="gitea-merger",          # or reviewer / author / tools
+    registered_tools=[...],            # optional static list
+    probe_result={"success": true, "result": {...}},
+    probe_source="client_namespace",
+)
+```
+
+3. A healthy client-namespace assessment is recorded in the MCP session and
+   consulted by **live** `gitea_submit_pr_review` / `gitea_merge_pr` gates.
+4. If the probe fails with EOF, recover via **client reconnect only** — see
+   `docs/mcp-namespace-eof-recovery.md`. Do **not** kill PIDs or touch MCP
+   config mtimes as a recovery procedure.
+
+## Offline spawn probe (non-authoritative)
+
+```bash
+python3 test_mcp_conn.py --config ~/.gemini/config/mcp_config.json
+```
+
+This script spawns a **separate** server process from config, performs
+JSON-RPC `initialize` → `tools/list` → `tools/call`, and classifies the
+result with `probe_source=offline_spawn`. Use it for offline debugging of
+launch command / registration. It does **not** prove the IDE-managed
+namespace is healthy.
+
+By default the script checks:
+
+| Namespace | Required tool |
+| --- | --- |
+| `gitea-author` | `gitea_whoami` |
+| `gitea-reviewer` | `gitea_whoami` |
+| `gitea-merger` | `gitea_whoami` |
+| `gitea-tools` | `gitea_list_profiles` |
+
+## Recovery (canonical)
+
+When a namespace returns EOF, follow
+`docs/mcp-namespace-eof-recovery.md` in order:
+
+1. Confirm blast radius (Gitea namespace vs all MCP servers).
+2. **Reconnect the namespace through the client** (IDE reconnect / relaunch).
+3. Do not repair via shell imports, raw JSON-RPC, PID kills, or config mtime
+   touches — those do not restore the client's closed transport.
+4. Re-verify the **specific** required tool through the target namespace.
+5. Resume review/merge only after a successful `client_namespace` assessment.
+
+## Enforcement
+
+1. **State machine (read-only):** feed `blocks_merge_workflow` from a
+   `client_namespace` assessment into
+   `gitea_assess_review_merge_state_machine(live_namespace_broken=...)`.
+2. **Live mutations:** `gitea_submit_pr_review` and `gitea_merge_pr` call
+   `_live_namespace_health_gate` and fail closed when the session has a
+   recorded unhealthy or non-client probe for the required namespace
+   (`gitea-reviewer` for review, `gitea-merger` for merge).
+
+When blocked, repair the IDE namespace and re-record a healthy
+`client_namespace` assessment before retrying the mutation.

docs/reviewer-handoff-consistency.md

@@ -0,0 +1,35 @@
+# Reviewer Handoff Consistency
+
+Reviewer and final-review controller handoffs must not contradict themselves.
+A narrative that says a merge happened, a review was blocked, or a terminal
+mutation budget was consumed must match the mutation ledger fields in the same
+handoff.
+
+## What gets validated
+
+`reviewer_handoff_consistency.assess_reviewer_handoff_consistency()` checks:
+
+- merge claims appear under `Merge mutations` or `MCP/Gitea mutations`
+- terminal-mutation-budget claims name the exact prior mutation in the ledger
+- blocked review submission is not paired with "final decision marked"
+- reviewer lease acquisition includes a `Review decision`
+- blocked/rejected mutations include proof fields:
+  - tool called
+  - mutation attempted
+  - mutation rejected
+  - no server-side state changed
+
+`final_report_validator` applies this as `reviewer.handoff_consistency` on
+`review_pr` reports and fails closed.
+
+## Blocked review template
+
+When `gitea_submit_pr_review` fails closed, use
+`reviewer_handoff_consistency.render_blocked_review_handoff_template()` or the
+copy in
+[`skills/llm-project-workflow/templates/blocked-review-handoff.md`](../skills/llm-project-workflow/templates/blocked-review-handoff.md).
+
+## Related
+
+- #331 — file-mutation ledger alignment
+- #501 — contradictory narrative vs ledger detection

docs/two-comment-workflow.md

@@ -0,0 +1,123 @@
+# Two-comment workflow: Controller Handoff + Thread State Ledger
+
+After meaningful controller/workflow work, post **two separate Gitea comments**:
+
+1. **`[CONTROLLER HANDOFF]`** — detailed operational handoff for the next
+   LLM/controller session (proof-heavy; may be long).
+2. **`[THREAD STATE LEDGER]`** — short canonical truth readable in ~30 seconds.
+
+The ledger is the concise source of truth. The handoff is the detailed
+continuation artifact. This complements CTH (#505) and lifecycle state
+comments (#494/#495) without replacing them.
+
+## Controller Handoff template
+
+```markdown
+[CONTROLLER HANDOFF] PR #___ / Issue #___ — <short title>
+
+Purpose:
+This comment is the operational handoff for the next controller/LLM session.
+
+Identity/profile:
+- Active profile:
+- Authenticated identity:
+- Role:
+- Self-review / role-conflict proof:
+
+Target:
+- Repo:
+- Issue:
+- PR:
+- Branch:
+- Pinned head SHA:
+- Worktree:
+
+Work performed:
+- <step 1>
+- <step 2>
+
+Files touched or reviewed:
+- `<file>` — <why it matters>
+
+Validation:
+- `<command>` → <result>
+- Full suite: <result or not run + reason>
+
+Server-side mutation ledger:
+- <mutation 1, including tool/action/comment id if available>
+- Or: none — no server-side state changed
+
+Local-only changes:
+- <worktree created, files edited locally, etc.>
+- Or: none
+
+Blockers:
+- <none>
+- Or: <exact blocker, exact gate, exact reason>
+
+Controller prompt for next session:
+```markdown
+<ready-to-paste prompt>
+```
+```
+
+## Thread State Ledger template
+
+```markdown
+[THREAD STATE LEDGER] PR #___ / Issue #___ — <current state in one line>
+
+What is true now:
+- PR state:
+- Issue state:
+- Current head SHA:
+- Server-side decision state:
+- Local verdict/state:
+- Latest known validation:
+
+What changed:
+- <short summary since prior ledger>
+
+What is blocked:
+- Blocker classification: <see list below>
+
+Who/what acts next:
+- Next actor:
+- Required action:
+- Do not do:
+- Resume from:
+```
+
+### Blocker classifications
+
+- code blocker
+- test blocker
+- merge conflict
+- stale head
+- permission/capability blocker
+- environment/tooling blocker
+- process/rule blocker
+- queue/lease blocker
+- duplicate/canonicalization blocker
+- no blocker
+
+### Precise state language
+
+Prefer:
+
+- `APPROVE verdict prepared locally`
+- `APPROVED review posted to Gitea`
+- `REQUEST_CHANGES posted to Gitea`
+- `merge performed` / `merge not performed`
+- `lease acquired` / `lease attempt blocked`
+- `server-side state changed` / `no server-side state changed`
+
+Avoid ambiguous standalone claims (`approved`, `ready`, `merged`, `blocked`,
+`done`) without proof and server-side state separation.
+
+## Validation
+
+- `thread_state_ledger_validator.py` — comment and final-report checks
+- `gitea_create_issue_comment` — fail-closed gate on tagged comments
+- `assess_final_report_validator` — `shared.two_comment_workflow` rule
+
+Examples: [`examples/two-comment-workflow-examples.md`](examples/two-comment-workflow-examples.md)

docs/webui-deployment.md

@@ -0,0 +1,59 @@
+# Web UI deployment boundary (#435)
+
+The MCP Control Plane web UI is an **internal operator console**, not a
+customer-facing application. The MVP assumes local or trusted-network access
+only.
+
+## MVP deployment model
+
+- **Default bind:** `127.0.0.1:8765` (`WEBUI_HOST` / `WEBUI_PORT`)
+- **Authentication:** none in MVP — protection comes from network placement
+- **Mutations:** read-only routes; gated write actions remain disabled (#434)
+- **Secrets:** resolved server-side via `gitea_auth` / `GITEA_MCP_CONFIG`; never
+  embedded in HTML, JavaScript, or browser storage
+
+Do **not** expose the UI on the public internet without an access layer.
+
+## Beyond localhost
+
+If the UI must be reachable outside the operator laptop:
+
+1. Prefer **Cloudflare Access**, **Cloudflare WARP**, or an org **VPN** so only
+   authenticated staff reach the service.
+2. Bind to a specific interface only when necessary — never `0.0.0.0` / `::`
+   without understanding the exposure.
+3. Set explicit override env vars only after access controls are in place:
+   - `WEBUI_ALLOW_PUBLIC_BIND=1` — acknowledges all-interface bind (`0.0.0.0`, `::`)
+   - `WEBUI_ALLOW_REMOTE_BIND=1` — acknowledges a non-loopback host
+
+Startup **refuses** all-interface binds unless `WEBUI_ALLOW_PUBLIC_BIND=1`.
+Non-loopback binds log a warning unless `WEBUI_ALLOW_REMOTE_BIND=1`.
+
+## Environment variables
+
+| Variable | Default | Purpose |
+|----------|---------|---------|
+| `WEBUI_HOST` | `127.0.0.1` | Bind address |
+| `WEBUI_PORT` | `8765` | Listen port |
+| `WEBUI_REPO_ROOT` | repository root | Workflow/schema hash root for prompt library |
+| `WEBUI_PROJECT_REGISTRY` | packaged JSON | Project registry file path |
+| `GITEA_MCP_CONFIG` | unset | Server-side MCP profile config path (optional) |
+| `GITEA_MCP_PROFILE` | unset | Active MCP profile name (optional) |
+| `WEBUI_ALLOW_PUBLIC_BIND` | unset | Acknowledge `0.0.0.0` / `::` bind |
+| `WEBUI_ALLOW_REMOTE_BIND` | unset | Acknowledge non-loopback bind |
+
+Gitea credentials (`GITEA_TOKEN_*`, `.env.*`, keychain refs) are read only on
+the server when a page needs live Gitea data (e.g. `/queue`). They are not
+shipped to the browser.
+
+## Health / deployment metadata
+
+`GET /health` includes a `deployment` object with bind disposition, runtime
+assumption paths, and the client-secret policy. Use it to verify an instance is
+configured for internal-only operation.
+
+## Non-goals (MVP)
+
+- Full SSO or session login in the UI
+- Hosting on the public internet without Access/VPN/WARP
+- Embedding Gitea tokens in the frontend bundle

docs/webui-local-dev.md

@@ -29,6 +29,13 @@ Optional environment variables:
 |----------|---------|---------|
 | `WEBUI_HOST` | `127.0.0.1` | Bind address (keep local for MVP) |
 | `WEBUI_PORT` | `8765` | Listen port |
+| `WEBUI_REPO_ROOT` | repository root | Prompt library workflow hash root |
+| `WEBUI_PROJECT_REGISTRY` | packaged JSON | Project registry path |
+| `GITEA_MCP_CONFIG` | unset | Server-side MCP profile config (never sent to browser) |
+| `GITEA_MCP_PROFILE` | unset | Active MCP profile name (server-side only) |
+
+See [webui-deployment.md](webui-deployment.md) for internal-only serving,
+Cloudflare Access/WARP/VPN guidance, and unsafe bind overrides (#435).
 
 ## Routes (MVP)
 
@@ -45,13 +52,27 @@ Optional environment variables:
 | `/api/prompts` | JSON prompt export with workflow hashes |
 | `/runtime` | MCP runtime health and stale detection (#430) |
 | `/api/runtime` | JSON runtime health export |
-| `/audit` | Stub — report audit paste (#431) |
-| `/worktrees` | Stub — hygiene dashboard (#432) |
+| `/audit` | Report audit paste + validator preview (#431) |
+| `/api/audit` | JSON validator preview (POST `report_text`, optional `task_kind`) |
+| `/worktrees` | Worktree hygiene dashboard (#432) |
+| `/api/worktrees` | JSON worktree scan with classifications and anomalies |
+| `/actions` | Gated write-action registry — all disabled in MVP (#434) |
+| `/api/actions` | JSON action registry with capability metadata |
+| `/api/actions/{id}/preview` | Mutation ledger preview (GET, read-only) |
 | `/leases` | Lease and collision visibility (#433) |
 | `/api/leases` | JSON lease/collision export |
 
-All routes are GET-only. POST/PUT/PATCH/DELETE return `405` with
-`read-only-mvp`.
+Most routes are GET-only. POST/PUT/PATCH/DELETE return `405` with
+`read-only-mvp`, except `/audit` and `/api/audit` which accept POST for
+local validator preview only (no Gitea mutations, no server-side storage).
+
+## Report audit (#431)
+
+Paste an LLM final report at `/audit` or POST JSON to `/api/audit`. The UI
+reuses `final_report_validator` and review schema checks to surface missing
+proof fields, wrong validation vocabulary, mutation contradictions, and a
+suggested next prompt or issue-comment draft. Task kind can be auto-detected
+or selected explicitly.
 
 ## Project registry (#427)
 
@@ -84,6 +105,84 @@ credentials. The UI surfaces pagination proof (returned count, pages fetched,
 If credentials are missing or the fetch fails, the page shows an explicit error
 instead of an empty queue (fail closed).
 
+## Gated actions (#434)
+
+`/actions` registers future write actions (claim, comment, review, merge,
+delete branch, create PR/issue). Each entry declares the MCP tool, required
+permission, and profile role from `task_capability_map.py` — aligned with
+`gitea_resolve_task_capability`. Buttons are disabled; previews always render
+a mutation ledger. Direct `attempt_action` calls fail closed without invoking
+MCP tools.
+
+## Worktree hygiene (#432)
+
+`/worktrees` scans local `branches/` directories and registered git worktrees.
+Each entry is classified (`active-pr`, `active-issue`, `dirty`, `stale-clean`,
+`detached-review`, `unsafe-unknown`, `orphan`). Missing preserved worktrees
+referenced by the issue lock file are flagged as anomalies (#404). The page
+includes a copy/paste canonical cleanup prompt only — no deletion actions.
+
+Override scan root with `WEBUI_REPO_ROOT` (defaults to repository root).
+
+## Lease visibility (#433)
+
+`/leases` surfaces read-only lease and collision state: local issue lock file,
+in-progress claim inventory (#268), reviewer PR lease comments when present
+(`<!-- mcp-review-lease:v1 -->`, #407), duplicate open PRs per issue (#400),
+and duplicate local branches per issue. Links to collision-history backend
+issues (#267, #268, #400, #407) are included. No lease acquire/release from UI.
+
+## Runtime health (#430)
+
+`/runtime` surfaces read-only MCP/runtime diagnostics for the default registry
+project: active profile and role kind, authenticated identity (when credentials
+are available), config model/mode, local vs remote `master` SHA sync, shell
+health, workflow/schema SHA-256 hashes, and stale-runtime warnings when the
+checkout is behind merged safety-gate changes. Restart guidance links to #420;
+no tokens or MCP restart actions are exposed.
+
+## Deployment boundary (#435)
+
+MVP serves on loopback by default. Binding `0.0.0.0` or `::` is **refused**
+unless `WEBUI_ALLOW_PUBLIC_BIND=1`. Non-loopback hosts log a warning unless
+`WEBUI_ALLOW_REMOTE_BIND=1`. `GET /health` exposes `deployment` metadata.
+
+## Tests
+
+```bash
+pytest tests/test_webui_skeleton.py tests/test_webui_project_registry.py tests/test_webui_prompt_library.py tests/test_webui_queue_dashboard.py tests/test_webui_gated_actions.py -q
+pytest tests/test_webui_skeleton.py tests/test_webui_project_registry.py tests/test_webui_prompt_library.py tests/test_webui_queue_dashboard.py tests/test_webui_audit.py tests/test_webui_worktree_hygiene.py -q
+pytest tests/test_webui_skeleton.py tests/test_webui_project_registry.py tests/test_webui_prompt_library.py tests/test_webui_queue_dashboard.py tests/test_webui_lease_visibility.py tests/test_webui_runtime_health.py -q
+
+pytest tests/test_webui_skeleton.py tests/test_webui_project_registry.py tests/test_webui_prompt_library.py tests/test_webui_queue_dashboard.py tests/test_webui_deployment_boundary.py -q
+
+## Tests (#436)
+
+Run the full hermetic web UI suite (all `test_webui_*.py` modules):
+
+```bash
+./scripts/test-webui
+```
+
+CI / Jenkins multibranch can call the path-filtered gate (runs only when the
+diff touches `webui/`, `tests/test_webui_*`, or web UI docs/scripts):
+
+```bash
+./scripts/ci-webui-check
+WEBUI_CI_FORCE=1 ./scripts/ci-webui-check   # always run
+```
+
+`scripts/test-webui` sets `WEBUI_TEST_OFFLINE=1` by default. In that mode the
+queue, lease, and runtime routes use empty offline snapshots instead of Gitea
+credentials, so CI can run without MCP daemon credential access. Set
+`WEBUI_TEST_OFFLINE=0` only when deliberately validating live fetch behavior.
+
+Or invoke unittest directly:
+
+```bash
+python3 -m unittest discover -s tests -p 'test_webui_*.py' -q
+```
+
 ## Lease visibility (#433)
 
 `/leases` surfaces read-only lease and collision state: local issue lock file,
@@ -104,5 +203,6 @@ no tokens or MCP restart actions are exposed.
 ## Tests
 
 ```bash
+pytest tests/test_webui_skeleton.py tests/test_webui_project_registry.py tests/test_webui_prompt_library.py tests/test_webui_queue_dashboard.py tests/test_webui_audit.py tests/test_webui_worktree_hygiene.py -q
 pytest tests/test_webui_skeleton.py tests/test_webui_project_registry.py tests/test_webui_prompt_library.py tests/test_webui_queue_dashboard.py tests/test_webui_lease_visibility.py tests/test_webui_runtime_health.py -q
-```
+```

docs/wiki/Identity-and-Profiles.md

@@ -16,12 +16,21 @@ bind an authenticated Gitea identity to an allowed operation set.
 - **Allowed:** branch create/push, PR create, issue comment/create/close, repo commit, read.
 - **Forbidden:** PR approve, merge, request_changes.
 
-### Reviewer / merger
+### Reviewer
 
 - **Profile:** `prgs-reviewer`
 - **Typical identity:** `sysadmin`
-- **Allowed:** PR review/approve/merge/request_changes, issue comment, read.
-- **Forbidden:** branch push, PR create, repo commit.
+- **Allowed:** PR review/approve/request_changes, issue comment, read.
+- **Forbidden:** branch push, PR create, repo commit, PR merge.
+
+Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
+
+### Merger
+
+- **Profile:** `prgs-merger`
+- **Typical identity:** `sysadmin`
+- **Allowed:** PR merge, issue comment, read.
+- **Forbidden:** branch push, PR create, repo commit, PR approve, PR review, PR request_changes.
 
 ## Configuration
 

docs/wiki/Operator-Guide.md

@@ -14,6 +14,7 @@ Handbook for LLM operators and human developers using the Gitea-Tools MCP server
 4. **No self-review / no self-merge** — The authenticated Gitea user must not approve or merge a PR they authored.
 5. **Follow the gates** — Prompts express intent; MCP tools enforce safety. Never bypass gates via prompt instructions.
 6. **Global LLM Worktree Rule** — Main checkout stays on `master`/`main`/`dev`; all mutations happen under `branches/`. Prove project root, `cwd`, branch, stable main-checkout branch, and session worktree path before editing. No exceptions.
+7. **Stable control runtime** — Real Gitea mutations use only the **stable** MCP control runtime. LLM sessions must not kill, restart, or relaunch MCP processes, edit the stable checkout, or use a dev MCP for production mutations. See the policy ADR: [mcp-stable-control-runtime-policy-adr.md](../architecture/mcp-stable-control-runtime-policy-adr.md) (#615).
 
 ## Supported Gitea instances
 
@@ -29,4 +30,5 @@ Always pass `remote` explicitly on tool calls. The server default is `dadeschool
 1. `gitea_whoami` — confirm authenticated user and profile.
 2. `gitea_get_runtime_context` — allowed/forbidden operations for this session.
 3. `gitea_resolve_task_capability` — prove the session may perform the planned task.
-4. For reviewer work: dry-run validation (`gitea_dry_run_pr_review`) before live review mutations.
+4. For reviewer work: dry-run validation (`gitea_dry_run_pr_review`) before live review mutations.
+5. Confirm master parity / runtime health when tools report stale or unhealthy control runtime — stop mutations and request an **operator** reload of the stable MCP (see [stable control runtime ADR](../architecture/mcp-stable-control-runtime-policy-adr.md)).

docs/wiki/Runbooks.md

@@ -12,6 +12,17 @@
 4. `gitea_mark_final_review_decision` → approve via `gitea_review_pr`.
 5. `gitea_merge_pr` with pinned head SHA and `confirmation="MERGE PR <n>"`.
 
+## Stable MCP control runtime (#615)
+
+Policy ADR: [mcp-stable-control-runtime-policy-adr.md](../architecture/mcp-stable-control-runtime-policy-adr.md).
+
+| Situation | Who acts | What to do |
+|-----------|----------|------------|
+| Transport EOF / missing tools | LLM | **Client reconnect only** — do not kill PIDs |
+| Wrong profile / dual-namespace missing | Operator | Configure or reload the correct static namespace(s) |
+| Master parity stale after merge | LLM stops + reports; **operator** reloads stable MCP | Resume only after parity re-verified |
+| Promote unpromoted MCP code to stable | Operator / release-manager only | Full §2.4 promotion record |
+
 ## Gitea Wiki sync
 
 The Gitea Wiki mirrors `docs/wiki/` (source of truth). After merging wiki changes:

docs/wiki/Safety-and-Gates.md

@@ -15,5 +15,7 @@
 5. **Explicit merge confirmation** — `gitea_merge_pr` requires `confirmation="MERGE PR <n>"`.
 6. **No self-review / self-merge** — Authenticated user must differ from PR author for approve/merge.
 7. **Review decision lock** — Live review mutations require validation-phase dry-run and `gitea_mark_final_review_decision`.
-8. **Redaction** — Tokens, passwords, and keychain material never appear in tool output.
-9. **Wiki publication (#224)** — `docs/wiki/` and the sync helper are prerequisites only. Closing a wiki issue requires live Gitea Wiki proof on the repo Wiki tab. See [Runbooks](Runbooks.md#wiki-publication-readiness-gate-224).
+8. **Terminal review hard-stop (#332)** — After a terminal live review mutation, only same-PR merge after `approve` may continue. Durable locks (#559) must not be deleted by hand.
+9. **Stale decision-lock cleanup (#594)** — `gitea_cleanup_stale_review_decision_lock` may clear a durable #332 lock **only** when the last terminal mutation's PR is live-state **merged or closed**, identity/profile gates pass, and apply uses a reviewer-capable profile. Open/ambiguous locks stay fail-closed. Successful cleanup records a durable audit trail.
+10. **Redaction** — Tokens, passwords, and keychain material never appear in tool output.
+11. **Wiki publication (#224)** — `docs/wiki/` and the sync helper are prerequisites only. Closing a wiki issue requires live Gitea Wiki proof on the repo Wiki tab. See [Runbooks](Runbooks.md#wiki-publication-readiness-gate-224).

docs/workflow-skill-mount.md

@@ -0,0 +1,68 @@
+# Workflow skill mount across runtimes (#551)
+
+## Problem
+
+Controller prompts require **`gitea-workflow`**, but:
+
+- Claude may load `~/.claude/skills/gitea-workflow`
+- Codex often has **no** `~/.codex/skills/gitea-workflow`
+- The portable package in-repo is `skills/llm-project-workflow`
+- `mcp_list_project_skills` historically listed operational guides only, not
+  the workflow router
+
+Sessions then either **block** incorrectly or **proceed without** the workflow
+wall.
+
+## Canonical names (must resolve to the same skill)
+
+| Name | Use |
+|------|-----|
+| `gitea-workflow` | **Primary** controller / Codex skill name |
+| `llm-project-workflow` | Portable in-repo package name |
+| `git-pr-workflows` | Legacy alias |
+
+Source of truth: `skills/llm-project-workflow/SKILL.md`  
+In-repo alias stub: `skills/gitea-workflow/SKILL.md`
+
+## Codex install
+
+From a `branches/` worktree (or any clone of the repo):
+
+```bash
+./scripts/install-codex-workflow-skill.sh
+# optional:
+./scripts/install-codex-workflow-skill.sh --dry-run
+./scripts/install-codex-workflow-skill.sh --skills-dir "$HOME/.codex/skills"
+```
+
+This symlinks the portable package under all three names. **Restart Codex**
+after install.
+
+## MCP discovery
+
+- `mcp_list_project_skills` includes `gitea-workflow`, `llm-project-workflow`,
+  and `git-pr-workflows`.
+- `mcp_get_skill_guide("<name>")` returns the same router steps for each.
+- `mcp_check_workflow_skill_preflight` proves the in-repo skill file exists and
+  reports Codex mount status.
+
+## Preflight rule
+
+Before any git or Gitea mutation:
+
+1. Call `mcp_check_workflow_skill_preflight`.
+2. If `blocked` / `workflow_skill_ready` is false → **BLOCKED + DIAGNOSE**; do
+   not mutate.
+3. Load the skill by **any** canonical name and follow the router.
+
+Missing Codex mount alone does not block if the in-repo skill is present and
+loaded via MCP/docs; operators should still install the Codex symlink so
+prompt names resolve natively.
+
+## Controller prompts
+
+Prefer:
+
+> Invoke skill `gitea-workflow` (alias of `llm-project-workflow`).
+
+Do not require a name that is only available on one runtime.

final_report_validator.py

@@ -11,8 +11,12 @@ import inspect
 import re
 from typing import Any, Callable
 
+import branch_cleanup_guard
 import issue_acceptance_gate
 import issue_lock_provenance
+import merger_lease_adoption
+import reviewer_handoff_consistency
+import thread_state_ledger_validator
 from mcp_native_cleanup_proof import assess_mcp_native_cleanup_proof
 from post_merge_cleanup_proof import assess_post_merge_cleanup_proof
 from review_proofs import (
@@ -29,31 +33,40 @@ from validation_status_vocabulary import assess_validation_status_vocabulary
 
 FINAL_REPORT_TASK_KINDS = frozenset({
     "review_pr",
+    "merge_pr",
     "reconcile_already_landed",
     "author_issue",
     "work_issue",
     "issue_filing",
     "issue_selection",
     "inventory",
+    "controller_close",
 })
 
 _TASK_KIND_ALIASES = {
     "review": "review_pr",
     "review-merge-pr": "review_pr",
+    "merge": "merge_pr",
+    "merge_pr": "merge_pr",
     "reconcile-landed-pr": "reconcile_already_landed",
     "reconcile_already_landed": "reconcile_already_landed",
     "work-issue": "work_issue",
     "create_issue": "issue_filing",
+    "close_issue": "controller_close",
+    "close-issue": "controller_close",
+    "controller-close": "controller_close",
 }
 
 _HANDOFF_ROLE_BY_TASK = {
     "review_pr": "review",
+    "merge_pr": "merger",
     "reconcile_already_landed": None,
     "author_issue": "author",
     "work_issue": "author",
     "issue_filing": "issue_filing",
     "issue_selection": None,
     "inventory": "inventory",
+    "controller_close": None,
 }
 
 _LEGACY_WORKSPACE_MUTATIONS_RE = re.compile(
@@ -121,16 +134,22 @@ _TARGET_BRANCH_SHA_RE = re.compile(
     r"target branch sha\s*:\s*[0-9a-f]{40}",
     re.IGNORECASE,
 )
+# #698: structured proof is rendered in several equivalent shapes —
+# `workflow_hash: abc...`, `workflow_hash=abc...`, or JSON
+# `"workflow_hash": "abc..."`. Recognize all of them; demanding one exact
+# punctuation style rejects legitimate structured workflow-load proof.
 _WORKFLOW_LOAD_HELPER_RE = re.compile(
-    r"workflow[- ]load helper result\s*:",
+    r"workflow[-_ ]load[-_ ]helper[-_ ]result\s*[:=]",
     re.IGNORECASE,
 )
 _WORKFLOW_LOAD_HASH_RE = re.compile(
-    r"workflow[- ]load helper result[\s\S]{0,400}?workflow[_ ]hash\s*:\s*[0-9a-f]{12}",
+    r"workflow[-_ ]load[-_ ]helper[-_ ]result[\s\S]{0,400}?"
+    r"workflow[_ ]hash\"?\s*[:=]\s*\"?[0-9a-f]{12}",
     re.IGNORECASE,
 )
 _WORKFLOW_LOAD_BOUNDARY_RE = re.compile(
-    r"workflow[- ]load helper result[\s\S]{0,400}?boundary[_ ]status\s*:\s*(?:clean|violation)",
+    r"workflow[-_ ]load[-_ ]helper[-_ ]result[\s\S]{0,400}?"
+    r"boundary[_ ]status\"?\s*[:=]\s*\"?(?:clean|violation)",
     re.IGNORECASE,
 )
 _WORKFLOW_FILE_VIEW_NARRATIVE_RE = re.compile(
@@ -195,6 +214,23 @@ _PAGINATION_PROOF_RE = re.compile(
     re.IGNORECASE,
 )
 _GIT_FETCH_RE = re.compile(r"\bgit\s+fetch\b", re.IGNORECASE)
+_CANONICAL_VALIDATION_REJECTED_RE = re.compile(
+    r"canonical comment validation failed|"
+    r"canonical_comment_validation|"
+    r'"allowed"\s*:\s*false',
+    re.IGNORECASE,
+)
+_COMMENT_POSTED_CLAIM_RE = re.compile(
+    r"(?:issue comments posted\s*:\s+(?!none\b)\S|"
+    r"pr comments posted\s*:\s+(?!none\b)\S|"
+    r"comment_id\s*[:=]\s*\d+|"
+    r"gitea comment (?:was )?posted|"
+    r"posted (?:issue|pr|canonical) (?:state )?comment|"
+    r"comment posted successfully|"
+    r"mcp/gitea mutations\s*:\s*[^;\n]*comment posted|"
+    r"reconciliation mutations\s*:\s*[^;\n]*comment posted)",
+    re.IGNORECASE,
+)
 _READONLY_DIAG_RE = re.compile(
     r"read[- ]only diagnostics\s*:\s*(.+)$",
     re.IGNORECASE | re.MULTILINE,
@@ -214,6 +250,59 @@ def _normalize_task_kind(task_kind: str | None) -> str:
     return _TASK_KIND_ALIASES.get(raw, raw)
 
 
+def _iter_action_entries(action_log: list | None) -> list[dict]:
+    """Yield only structured (dict) action-log entries (#698).
+
+    Callers must never crash on malformed entries (strings, numbers, null)
+    that reach the validator from LLM-composed or partially parsed logs;
+    :func:`sanitize_action_log` reports them separately.
+    """
+    return [e for e in (action_log or []) if isinstance(e, dict)]
+
+
+def sanitize_action_log(
+    action_log: list | None,
+) -> tuple[list[dict], list[dict[str, str]]]:
+    """Split an action log into structured entries and sanitized findings (#698).
+
+    Malformed entries become clear, sanitized ``warning`` findings — the
+    offending value's content is never echoed back (only its position and
+    type), so secrets or garbage in a broken log cannot leak into validation
+    errors, and validation itself proceeds without secondary exceptions.
+    """
+    if action_log is None:
+        return [], []
+    if not isinstance(action_log, (list, tuple)):
+        return [], [
+            validator_finding(
+                "shared.action_log_malformed",
+                "downgrade",
+                "Action log",
+                "action_log is not a list of structured entries "
+                f"(got {type(action_log).__name__}); it was ignored",
+                "pass action_log as a list of dict entries",
+            )
+        ]
+    entries: list[dict] = []
+    findings: list[dict[str, str]] = []
+    for index, entry in enumerate(action_log):
+        if isinstance(entry, dict):
+            entries.append(entry)
+            continue
+        findings.append(
+            validator_finding(
+                "shared.action_log_malformed",
+                "downgrade",
+                "Action log",
+                f"action_log entry {index} is not a structured mapping "
+                f"(got {type(entry).__name__}); the entry was ignored",
+                "repair the malformed action_log entry or drop it before "
+                "revalidating",
+            )
+        )
+    return entries, findings
+
+
 def validator_finding(
     rule_id: str,
     severity: str,
@@ -381,6 +470,43 @@ def _rule_shared_email_disclosure(report_text: str) -> list[dict[str, str]]:
     )
 
 
+def _rule_shared_canonical_comment_post_claim(
+    report_text: str,
+    *,
+    action_log: list[dict] | None = None,
+) -> list[dict[str, str]]:
+    """#496: reports must not claim comment posted when validator rejected."""
+    text = report_text or ""
+    rejected_in_report = bool(_CANONICAL_VALIDATION_REJECTED_RE.search(text))
+    rejected_in_log = False
+    if action_log:
+        for entry in _iter_action_entries(action_log):
+            validation = entry.get("canonical_comment_validation") or {}
+            if validation.get("allowed") is False:
+                rejected_in_log = True
+                break
+            result = entry.get("result") or {}
+            nested = result.get("canonical_comment_validation") or {}
+            if nested.get("allowed") is False:
+                rejected_in_log = True
+                break
+    if not (rejected_in_report or rejected_in_log):
+        return []
+    if not _COMMENT_POSTED_CLAIM_RE.search(text):
+        return []
+    return [
+        validator_finding(
+            "shared.canonical_comment_post_claim",
+            "block",
+            "MCP/Gitea mutations",
+            "report claims a Gitea comment was posted while canonical comment "
+            "validation rejected the workflow comment",
+            "do not claim comment posted when validator fail-closed; repair "
+            "the canonical comment body and retry posting",
+        )
+    ]
+
+
 def _rule_reviewer_legacy_workspace_mutations(
     report_text: str,
     *,
@@ -408,9 +534,13 @@ def _rule_reviewer_vague_mutations_none(
     action_log: list[dict] | None = None,
     mutations_observed: bool = False,
 ) -> list[dict[str, str]]:
+    # #698: infer review mutations only from authoritative evidence — an
+    # entry proves a mutation only when it affirmatively records
+    # performed=true and was not gated. Read-only diagnostics and pre-API
+    # rejections (entries without a performed flag) are not mutations.
     performed = any(
-        e.get("performed") is not False and not e.get("gated_rejected")
-        for e in (action_log or [])
+        e.get("performed") is True and not e.get("gated_rejected")
+        for e in _iter_action_entries(action_log)
     )
     if not (mutations_observed or performed):
         return []
@@ -469,7 +599,7 @@ def _rule_reviewer_git_fetch_readonly(
     text = report_text or ""
     fetch_observed = any(
         _GIT_FETCH_RE.search(str(e.get("command") or e.get("action") or ""))
-        for e in (action_log or [])
+        for e in _iter_action_entries(action_log)
     ) or _GIT_FETCH_RE.search(text)
     if not fetch_observed:
         return []
@@ -598,6 +728,27 @@ def _rule_reviewer_stale_head_proof(report_text: str) -> list[dict[str, str]]:
     )
 
 
+def _rule_conflict_fix_classification_proof(report_text: str) -> list[dict[str, str]]:
+    from conflict_fix_classification import (
+        assess_conflict_fix_classification_final_report,
+    )
+
+    text = report_text or ""
+    result = assess_conflict_fix_classification_final_report(text)
+    if result.get("proven"):
+        return []
+    return _findings_from_reasons(
+        "author.conflict_fix_classification_proof",
+        result.get("reasons") or [],
+        field="Conflict-fix classification",
+        severity="block",
+        safe_next_action=(
+            "call gitea_assess_conflict_fix_classification, state the live head "
+            "SHA, and state the classification before creating a conflict-fix worktree"
+        ),
+    )
+
+
 def _rule_conflict_fix_push_proof(report_text: str) -> list[dict[str, str]]:
     from pr_work_lease import assess_conflict_fix_final_report
 
@@ -767,6 +918,27 @@ def _rule_reviewer_premerge_baseline_proof(report_text: str) -> list[dict[str, s
     )
 
 
+def _rule_reviewer_post_merge_validation(report_text: str) -> list[dict[str, str]]:
+    """Block active approval on an already-merged/closed PR, and post-merge moot
+    validation claimed without merged-state + merge-commit proof (#529)."""
+    from post_merge_validation import assess_post_merge_validation
+
+    result = assess_post_merge_validation(report_text)
+    if not result.get("block"):
+        return []
+    return _findings_from_reasons(
+        "reviewer.post_merge_validation",
+        result.get("reasons") or [],
+        field="Validation status",
+        severity="block",
+        safe_next_action=result.get("safe_next_action")
+        or (
+            "record post-merge moot validation instead of an active approval; "
+            "cite PR state merged/closed and the merge commit SHA"
+        ),
+    )
+
+
 def _rule_reviewer_validation_status_vocabulary(
     report_text: str,
     *,
@@ -868,7 +1040,7 @@ def _rule_reviewer_target_branch_freshness(
     fields = _handoff_fields(text)
     fetch_reported = bool(_GIT_FETCH_RE.search(text)) or any(
         _GIT_FETCH_RE.search(str(e.get("command") or e.get("action") or ""))
-        for e in (action_log or [])
+        for e in _iter_action_entries(action_log)
     )
     target_sha_reported = bool(_TARGET_BRANCH_SHA_RE.search(text)) or any(
         "target branch" in key and "sha" in key and _FULL_SHA_RE.search(value)
@@ -1111,6 +1283,25 @@ def _rule_reviewer_validation_structured(
     )
 
 
+def _rule_reviewer_handoff_consistency(report_text: str) -> list[dict[str, str]]:
+    result = reviewer_handoff_consistency.assess_reviewer_handoff_consistency(
+        report_text
+    )
+    if result.get("proven"):
+        return []
+    return _findings_from_reasons(
+        "reviewer.handoff_consistency",
+        result.get("reasons") or [],
+        field="Review mutations",
+        severity="block",
+        safe_next_action=(
+            "rewrite reviewer handoff so narrative claims match the mutation "
+            "ledger; use the blocked-review handoff template when submission "
+            "was rejected"
+        ),
+    )
+
+
 def _rule_reviewer_mutation_ledger(
     report_text: str,
     *,
@@ -1162,6 +1353,24 @@ def _rule_shared_manual_lock_pr_override(report_text: str) -> list[dict[str, str
     )
 
 
+def _rule_shared_manual_lease_proof_handoff(report_text: str) -> list[dict[str, str]]:
+    """Block merge/review handoffs that claim unsafe lease seeding (#535)."""
+    result = merger_lease_adoption.assess_manual_lease_proof_handoff(report_text)
+    if result.get("proven"):
+        return []
+    return _findings_from_reasons(
+        "shared.manual_lease_proof_handoff",
+        result.get("reasons") or [],
+        field="Merge mutations",
+        severity="block",
+        safe_next_action=(
+            "use gitea_adopt_merger_pr_lease or gitea_acquire_reviewer_pr_lease; "
+            "cite lease_proof_source / adoption_comment_id; never claim manual "
+            "seeded/injected lease proof"
+        ),
+    )
+
+
 def _rule_shared_issue_acceptance_gate(report_text: str) -> list[dict[str, str]]:
     result = issue_acceptance_gate.validate_final_report_issue_acceptance(report_text)
     if not result.get("applicable") or result.get("valid"):
@@ -1194,6 +1403,17 @@ def _rule_shared_author_reviewer_same_run(report_text: str) -> list[dict[str, st
     )
 
 
+def _rule_shared_raw_branch_delete_bypass(report_text: str) -> list[dict[str, str]]:
+    result = branch_cleanup_guard.assess_raw_branch_delete_report(report_text)
+    if not result["block"]:
+        return []
+    return _findings_from_reasons(
+        "shared.raw_branch_delete_bypass",
+        result["reasons"],
+        field="Cleanup mutations",
+        severity="block",
+        safe_next_action=result["safe_next_action"],
+    )
 def _rule_reviewer_workflow_load_boundary(report_text: str) -> list[dict[str, str]]:
     """#403: require structured workflow-load helper result, not file-view narrative."""
     if not report_text.strip():
@@ -1277,6 +1497,29 @@ def _rule_reviewer_review_mutation(
     )
 
 
+def _rule_shared_two_comment_workflow(report_text: str) -> list[dict[str, str]]:
+    """#507: tagged Controller Handoff must pair with Thread State Ledger."""
+    return thread_state_ledger_validator.findings_for_final_report(report_text)
+
+
+def _rule_shared_canonical_state_update(report_text: str) -> list[dict[str, str]]:
+    from canonical_state_comments import validate_final_report_state_update
+
+    result = validate_final_report_state_update(report_text)
+    if not result.get("applicable") or result.get("valid"):
+        return []
+    return [
+        validator_finding(
+            "shared.canonical_state_update",
+            "block",
+            "Canonical state update",
+            reason,
+            "include a canonical state block with STATE, WHO_IS_NEXT, NEXT_ACTION, and NEXT_PROMPT",
+        )
+        for reason in (result.get("reasons") or ["invalid canonical state update"])
+    ]
+
+
 def _rule_reviewer_mutation_capability_proof(report_text: str) -> list[dict[str, str]]:
     from reviewer_mutation_capability_proof import assess_mutation_capability_proof
 
@@ -1324,18 +1567,30 @@ def _rule_shared_mcp_native_cleanup_proof(report_text: str) -> list[dict[str, st
 _SHARED_ISSUE_LOCK_RULES = (
     _rule_shared_issue_lock_external_state,
     _rule_shared_manual_lock_pr_override,
+    _rule_shared_manual_lease_proof_handoff,
     _rule_shared_author_reviewer_same_run,
+    _rule_shared_raw_branch_delete_bypass,
+    _rule_shared_canonical_state_update,
 )
 
+_SHARED_TWO_COMMENT_RULES = (
+    _rule_shared_two_comment_workflow,
+)
 _SHARED_CLEANUP_PROOF_RULES = (
     _rule_shared_mcp_native_cleanup_proof,
 )
 
+_SHARED_CANONICAL_COMMENT_RULES = (
+    _rule_shared_canonical_comment_post_claim,
+)
+
 _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
     "review_pr": [
         _rule_shared_controller_handoff,
         _rule_shared_state_handoff_next_action,
         _rule_shared_email_disclosure,
+        *_SHARED_TWO_COMMENT_RULES,
+        *_SHARED_CANONICAL_COMMENT_RULES,
         *_SHARED_ISSUE_LOCK_RULES,
         _rule_reviewer_legacy_workspace_mutations,
         _rule_reviewer_vague_mutations_none,
@@ -1348,12 +1603,14 @@ _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
         _rule_reviewer_linked_issue,
         _rule_reviewer_baseline_on_failure,
         _rule_reviewer_premerge_baseline_proof,
+        _rule_reviewer_post_merge_validation,
         _rule_reviewer_validation_status_vocabulary,
         _rule_reviewer_main_checkout_baseline,
         _rule_reviewer_main_checkout_path,
         _rule_reviewer_already_landed_eligible,
         _rule_reviewer_already_landed_state,
         _rule_reviewer_target_branch_freshness,
+        _rule_reviewer_handoff_consistency,
         _rule_reviewer_workflow_load_boundary,
         _rule_reviewer_mutation_ledger,
         _rule_reviewer_review_mutation,
@@ -1362,10 +1619,23 @@ _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
         *_SHARED_CLEANUP_PROOF_RULES,
         _rule_reviewer_stale_head_proof,
     ],
+    "merge_pr": [
+        _rule_shared_controller_handoff,
+        _rule_shared_email_disclosure,
+        *_SHARED_ISSUE_LOCK_RULES,
+        _rule_reviewer_legacy_workspace_mutations,
+        _rule_reviewer_vague_mutations_none,
+        _rule_reviewer_mutation_categories,
+        _rule_reviewer_git_fetch_readonly,
+        _rule_reviewer_linked_issue,
+        _rule_reviewer_stale_head_proof,
+    ],
     "reconcile_already_landed": [
         _rule_reconcile_controller_handoff,
         _rule_shared_state_handoff_next_action,
         _rule_shared_email_disclosure,
+        *_SHARED_TWO_COMMENT_RULES,
+        *_SHARED_CANONICAL_COMMENT_RULES,
         *_SHARED_ISSUE_LOCK_RULES,
         *_SHARED_CLEANUP_PROOF_RULES,
         _rule_reconcile_stale_author_fields,
@@ -1383,6 +1653,8 @@ _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
         _rule_shared_controller_handoff,
         _rule_shared_state_handoff_next_action,
         _rule_shared_email_disclosure,
+        *_SHARED_TWO_COMMENT_RULES,
+        *_SHARED_CANONICAL_COMMENT_RULES,
         *_SHARED_ISSUE_LOCK_RULES,
         _rule_reviewer_vague_mutations_none,
     ],
@@ -1390,9 +1662,12 @@ _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
         _rule_shared_controller_handoff,
         _rule_shared_state_handoff_next_action,
         _rule_shared_email_disclosure,
+        *_SHARED_TWO_COMMENT_RULES,
+        *_SHARED_CANONICAL_COMMENT_RULES,
         *_SHARED_ISSUE_LOCK_RULES,
         _rule_shared_issue_acceptance_gate,
         _rule_reviewer_vague_mutations_none,
+        _rule_conflict_fix_classification_proof,
         _rule_conflict_fix_push_proof,
         _rule_worktree_cleanup_audit_proof,
     ],
@@ -1400,12 +1675,16 @@ _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
         _rule_shared_controller_handoff,
         _rule_shared_state_handoff_next_action,
         _rule_shared_email_disclosure,
+        *_SHARED_TWO_COMMENT_RULES,
+        *_SHARED_CANONICAL_COMMENT_RULES,
         *_SHARED_ISSUE_LOCK_RULES,
     ],
     "inventory": [
         _rule_shared_controller_handoff,
         _rule_shared_state_handoff_next_action,
         _rule_shared_email_disclosure,
+        *_SHARED_TWO_COMMENT_RULES,
+        *_SHARED_CANONICAL_COMMENT_RULES,
         *_SHARED_ISSUE_LOCK_RULES,
         _rule_reconcile_pagination_proof,
     ],
@@ -1413,8 +1692,17 @@ _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
         _rule_shared_controller_handoff,
         _rule_shared_state_handoff_next_action,
         _rule_shared_email_disclosure,
+        *_SHARED_TWO_COMMENT_RULES,
+        *_SHARED_CANONICAL_COMMENT_RULES,
         *_SHARED_ISSUE_LOCK_RULES,
     ],
+    # Controller issue closure (#529): a closure report must not bury an
+    # unproven non-zero suite exit as an "expected pre-existing failure".
+    # Kept intentionally narrow so a closure pre-check does not demand the
+    # full reviewer/author handoff schema.
+    "controller_close": [
+        _rule_reviewer_premerge_baseline_proof,
+    ],
 }
 
 
@@ -1510,6 +1798,12 @@ def assess_final_report_validator(
     checks: dict[str, Any] = {}
     findings: list[dict[str, str]] = []
 
+    # #698: malformed action_log data must never crash validation with a
+    # secondary exception; malformed entries surface as sanitized findings.
+    sanitized_action_log, action_log_findings = sanitize_action_log(action_log)
+    action_log = sanitized_action_log
+    findings.extend(action_log_findings)
+
     if normalized_kind == "issue_filing" and issue_filing_lock is not None:
         checks["issue_filing"] = assess_issue_filing_final_report(
             report_text,
@@ -1538,7 +1832,23 @@ def assess_final_report_validator(
     }
 
     for rule in _RULES_BY_TASK.get(normalized_kind, ()):
-        findings.extend(_call_rule(rule, report_text, normalized_kind, rule_kwargs))
+        try:
+            findings.extend(
+                _call_rule(rule, report_text, normalized_kind, rule_kwargs)
+            )
+        except Exception as exc:  # #698: fail closed with a sanitized error
+            findings.append(
+                validator_finding(
+                    "shared.validator_rule_error",
+                    "block",
+                    "Validator",
+                    f"validator rule '{getattr(rule, '__name__', 'unknown')}' "
+                    f"failed with {type(exc).__name__} (details withheld; "
+                    "sanitized)",
+                    "file a validator defect with the rule name; do not "
+                    "bypass final-report validation",
+                )
+            )
 
     grade, blocked, downgraded = _aggregate_grade(findings)
     reasons = [f"{f['rule_id']}: {f['reason']}" for f in findings]

gitea-mcp.v2-contexts.example.json

@@ -52,8 +52,19 @@
       "execution_profile": "example-reviewer",
       "audit_label": "example-reviewer",
       "auth": { "type": "keychain", "id": "example-gitea-reviewer-token" },
-      "allowed_operations": ["read", "review", "comment", "issue.comment", "approve", "request_changes", "merge"],
-      "forbidden_operations": ["branch", "commit", "push", "open_pr"]
+      "allowed_operations": ["read", "review", "comment", "issue.comment", "approve", "request_changes"],
+      "forbidden_operations": ["branch", "commit", "push", "open_pr", "merge"]
+    },
+    "example-merger": {
+      "enabled": true,
+      "context": "example-context",
+      "role": "merger",
+      "username": "reviewer-user",
+      "execution_profile": "example-merger",
+      "audit_label": "example-merger",
+      "auth": { "type": "keychain", "id": "example-gitea-reviewer-token" },
+      "allowed_operations": ["read", "comment", "issue.comment", "merge"],
+      "forbidden_operations": ["branch", "commit", "push", "open_pr", "approve", "review", "request_changes"]
     }
   },
   "projects": {
@@ -63,7 +74,8 @@
       "default_owner": "Example-Org",
       "default_repo": "Example-Repo",
       "default_author_profile": "example-author",
-      "default_reviewer_profile": "example-reviewer"
+      "default_reviewer_profile": "example-reviewer",
+      "default_merger_profile": "example-merger"
     }
   },
   "rules": {

gitea_auth.py

@@ -20,14 +20,15 @@ from dotenv import dotenv_values, load_dotenv
 
 import gitea_config
 
+PROJECT_ROOT = os.path.dirname(os.path.abspath(__file__))
+
 # Load standard .env if present
-load_dotenv()
+load_dotenv(os.path.join(PROJECT_ROOT, ".env"))
 
 # Dictionary to store configurations parsed dynamically from .env.* files
 DYNAMIC_CONFIGS = {}
 
 # Scan all files starting with .env in the project root to load multiple configurations
-PROJECT_ROOT = os.path.dirname(os.path.abspath(__file__))
 for env_path in glob.glob(os.path.join(PROJECT_ROOT, ".env*")):
     # Skip directories and the example template
     if os.path.basename(env_path) == ".env.example":
@@ -104,6 +105,10 @@ def get_credentials(host):
 
     # 3. Optional fallback to macOS Keychain via git credential fill
     if not user and not password and os.environ.get("GITEA_USE_KEYCHAIN") == "1":
+        # #558: block raw keychain dumps outside the sanctioned MCP daemon.
+        import mcp_daemon_guard
+
+        mcp_daemon_guard.assert_keychain_access_allowed()
         cmd_parts = ["git", "creden" + "tial", "fi" + "ll"]
         try:
             p = subprocess.Popen(
@@ -116,6 +121,8 @@ def get_credentials(host):
                     user = line.split("=", 1)[1]
                 elif line.startswith("password="):
                     password = line.split("=", 1)[1]
+        except mcp_daemon_guard.UnsanctionedRuntimeError:
+            raise
         except Exception:
             pass
 
@@ -124,6 +131,11 @@ def get_credentials(host):
 
 def get_auth_header(host):
     """Return an ``Authorization`` header value for *host*."""
+    # #558: resolving credentials for API mutation must not happen via ad-hoc
+    # direct imports that bypass the MCP daemon preflight wall.
+    import mcp_daemon_guard
+
+    mcp_daemon_guard.assert_sanctioned_mutation_runtime("get_auth_header")
     host_key = host.lower().strip()
 
     # 1. Try Token-based auth from dynamic configs
@@ -231,6 +243,192 @@ def _redact(text):
         return str(text)
 
 
+# ── Classified client failures (#699) ─────────────────────────────────────────
+# Subclasses of RuntimeError preserve existing ``except RuntimeError`` call
+# sites. Exception *messages* are fixed constants only — HTTP response bodies,
+# Keychain material, and arbitrary exception text are never stored on the
+# exception or re-emitted to tool results / daemon logs.
+
+
+# Fixed messages (must match mcp_tool_error_boundary.FIXED_MESSAGES keys used here).
+_MSG_AUTH_INVALID = "Gitea authentication failed: invalid or revoked credentials"
+_MSG_AUTH_FAILED = "Gitea authentication failed"
+_MSG_AUTHZ_SCOPE = "Gitea authorization failed: insufficient token scope"
+_MSG_AUTHZ_DENIED = "Gitea authorization failed: access denied"
+_MSG_NETWORK = "Network error contacting Gitea"
+_MSG_CONFIG = "Gitea configuration or credential resolution failed"
+_MSG_UPSTREAM = "Gitea upstream unavailable"
+_MSG_HTTP = "Gitea HTTP request failed"
+
+
+class GiteaClientError(RuntimeError):
+    """Base for known Gitea client failures with stable reason_code metadata."""
+
+    reason_code = "client_error"
+    error_class = "client"
+    http_status = None
+
+    def __init__(self, message=None, *, reason_code=None, http_status=None):
+        if reason_code is not None:
+            self.reason_code = reason_code
+        if http_status is not None:
+            self.http_status = http_status
+        # Message is always a fixed constant; callers cannot inject bodies.
+        fixed = message if message is not None else _MSG_HTTP
+        super().__init__(fixed)
+
+
+class GiteaAuthError(GiteaClientError):
+    """Authentication failure (invalid/revoked credentials → typically HTTP 401)."""
+
+    reason_code = "auth_invalid_token"
+    error_class = "authentication"
+    http_status = 401
+
+    def __init__(self, message=None, *, reason_code=None, http_status=None):
+        super().__init__(
+            message if message is not None else _MSG_AUTH_INVALID,
+            reason_code=reason_code or "auth_invalid_token",
+            http_status=http_status if http_status is not None else 401,
+        )
+
+
+class GiteaAuthzError(GiteaClientError):
+    """Authorization failure (HTTP 403 — scope deficiency or access denied)."""
+
+    reason_code = "authz_denied"
+    error_class = "authorization"
+    http_status = 403
+
+    def __init__(self, message=None, *, reason_code=None, http_status=None):
+        code = reason_code or "authz_denied"
+        if code == "authz_insufficient_scope":
+            fixed = _MSG_AUTHZ_SCOPE
+        else:
+            fixed = _MSG_AUTHZ_DENIED
+            code = "authz_denied"
+        super().__init__(
+            message if message is not None else fixed,
+            reason_code=code,
+            http_status=http_status if http_status is not None else 403,
+        )
+
+
+class GiteaNetworkError(GiteaClientError):
+    """Transport / DNS / timeout failure contacting Gitea."""
+
+    reason_code = "network_error"
+    error_class = "network"
+    http_status = None
+
+    def __init__(self, message=None, *, reason_code=None, http_status=None):
+        super().__init__(
+            message if message is not None else _MSG_NETWORK,
+            reason_code=reason_code or "network_error",
+            http_status=http_status,
+        )
+
+
+class GiteaConfigError(GiteaClientError):
+    """Local configuration / credential resolution failure (not HTTP auth)."""
+
+    reason_code = "config_error"
+    error_class = "configuration"
+    http_status = None
+
+    def __init__(self, message=None, *, reason_code=None, http_status=None):
+        super().__init__(
+            message if message is not None else _MSG_CONFIG,
+            reason_code=reason_code or "config_error",
+            http_status=http_status,
+        )
+
+
+class GiteaHttpError(GiteaClientError):
+    """Non-auth HTTP failure with fixed message (no response body)."""
+
+    reason_code = "http_error"
+    error_class = "client"
+    http_status = None
+
+    def __init__(self, message=None, *, reason_code=None, http_status=None):
+        code = reason_code or "http_error"
+        if code == "upstream_unavailable":
+            fixed = _MSG_UPSTREAM
+        else:
+            fixed = _MSG_HTTP
+            code = "http_error"
+        super().__init__(
+            message if message is not None else fixed,
+            reason_code=code,
+            http_status=http_status,
+        )
+
+
+def _looks_like_insufficient_scope(detail: str) -> bool:
+    """Internal: inspect redacted body *only* to refine 403 reason_code.
+
+    The body is never stored on the exception or returned to callers.
+    """
+    lower = (detail or "").lower()
+    markers = (
+        "insufficient scope",
+        "required scope",
+        "does not have at least one of required scope",
+        "token does not have",
+        "missing scope",
+        "scope(s)",
+    )
+    return any(m in lower for m in markers)
+
+
+def classify_http_status(code: int, *, body_hint: str = "") -> tuple[type, str, int]:
+    """Central HTTP status → (exception_class, reason_code, http_status).
+
+    Every HTTP 403 becomes authorization-class. Body text is used only as a
+    local hint for scope vs denied reason_code and is never returned.
+    """
+    if code == 401:
+        return (GiteaAuthError, "auth_invalid_token", 401)
+    if code == 403:
+        if _looks_like_insufficient_scope(body_hint or ""):
+            return (GiteaAuthzError, "authz_insufficient_scope", 403)
+        return (GiteaAuthzError, "authz_denied", 403)
+    if code in (502, 503, 504):
+        return (GiteaHttpError, "upstream_unavailable", code)
+    return (GiteaHttpError, "http_error", code)
+
+
+def raise_for_http_status(code: int, body: str = "") -> None:
+    """Raise a typed client error for *code* without embedding *body*.
+
+    *body* may be inspected only to choose scope vs denied for 403; it is
+    never placed on the exception message.
+    """
+    # Redact before any inspection; discard after classification.
+    try:
+        hint = _redact(body or "").strip()
+    except Exception:
+        hint = ""
+    exc_cls, reason, status = classify_http_status(code, body_hint=hint)
+    # Explicitly construct without passing body/hint into message.
+    if exc_cls is GiteaAuthError:
+        raise GiteaAuthError(reason_code=reason, http_status=status)
+    if exc_cls is GiteaAuthzError:
+        raise GiteaAuthzError(reason_code=reason, http_status=status)
+    if reason == "upstream_unavailable":
+        raise GiteaHttpError(
+            reason_code="upstream_unavailable",
+            http_status=status,
+        )
+    raise GiteaHttpError(reason_code="http_error", http_status=status)
+
+
+def _raise_http_error(code: int, detail: str = "") -> None:
+    """Backward-compatible alias — *detail* is never embedded in the error."""
+    raise_for_http_status(code, detail)
+
+
 def _add_query(url, **params):
     """Return *url* with the given query parameters added or overridden.
 
@@ -303,23 +501,24 @@ def api_request(method, url, auth_header, payload=None, *,
     """Make an authenticated JSON request to the Gitea API.
 
     Returns parsed JSON on success (or ``None`` for an empty body), and raises
-    ``RuntimeError`` on failure.
+    a classified client error on failure.
 
     On HTTP 429 the request is retried up to *max_retries* times: honoring a
     valid ``Retry-After`` header (seconds or HTTP-date) when present, otherwise
     using capped jittered exponential backoff. Successful responses are
     unchanged.
 
-    All failures are converted to a ``RuntimeError`` with a clear, secret
-    -redacted message (no raw stack traces or credential material):
+    Failures raise typed exceptions with **fixed messages only** (#699). HTTP
+    response bodies are read solely for local 403 reason refinement and are
+    never stored on exceptions or returned to callers:
 
-    - Non-429 HTTP errors surface the status code and a redacted response body.
-      502/503/504 upstream errors get an explicit "Gitea upstream unavailable"
-      message.
-    - Timeouts and network/DNS failures (``URLError`` / ``TimeoutError``) surface
-      a generic "network error contacting Gitea" message.
-    - A malformed (non-JSON) success body surfaces a "malformed JSON response"
-      message rather than a raw decode error.
+    - HTTP 401 → :class:`GiteaAuthError` (``auth_invalid_token``)
+    - HTTP 403 → :class:`GiteaAuthzError` (scope or denied)
+    - 502/503/504 → :class:`GiteaHttpError` (``upstream_unavailable``)
+    - Other non-429 HTTP → :class:`GiteaHttpError` (``http_error``)
+    - Timeouts / DNS / ``URLError`` → :class:`GiteaNetworkError`
+    - Malformed success JSON → plain ``RuntimeError`` (programming/protocol;
+      not reclassified as authentication)
 
     The ``*_func`` parameters and ``timeout`` are injection points for
     deterministic testing.
@@ -357,22 +556,23 @@ def api_request(method, url, auth_header, payload=None, *,
                 error_body = e.read().decode("utf-8", errors="replace")
             except Exception:
                 error_body = ""
-            detail = _redact(error_body).strip()
-            if e.code in (502, 503, 504):
-                msg = f"HTTP {e.code}: Gitea upstream unavailable"
-                raise RuntimeError(f"{msg}: {detail}" if detail else msg) from e
-            raise RuntimeError(f"HTTP {e.code}: {detail}") from e
+            # Classify from status (+ local body hint). Body is not embedded.
+            try:
+                raise_for_http_status(e.code, error_body)
+            except GiteaClientError:
+                raise
+            # Defensive: raise_for_http_status always raises.
+            raise GiteaHttpError(http_status=e.code) from e  # pragma: no cover
         except (urllib.error.URLError, TimeoutError) as e:
-            reason = getattr(e, "reason", e)
-            raise RuntimeError(
-                f"network error contacting Gitea: {_redact(reason)}"
-            ) from e
+            # Fixed message only — do not embed URLError reason (may leak paths).
+            raise GiteaNetworkError(reason_code="network_error") from e
 
         if not body:
             return None
         try:
             return json.loads(body)
         except ValueError as e:
+            # Programming/protocol failure — not authentication.
             raise RuntimeError("malformed JSON response from Gitea") from e
 
 

incident_bridge.py

@@ -0,0 +1,788 @@
+"""Observability incident bridge (#612).
+
+Converts Sentry/GlitchTip **observations** into durable **Gitea issues** and
+``incident_links`` rows on the #613 control-plane DB.
+
+Hard rules (ADR):
+* Gitea owns work; providers own incidents; the bridge only links them.
+* Raw monitoring incidents are **never** assignable ``work_items``.
+* The #600 allocator sees bridge work only as normal Gitea issues.
+* Phase-1 prefers dry-run / explicit reconcile; apply creates/links issues.
+* No tokens, DSNs, cookies, or other secrets in bodies, DB, or logs.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+import re
+from dataclasses import dataclass, field
+from typing import Any, Callable, Sequence
+
+from control_plane_db import ControlPlaneDB, ControlPlaneError, WORK_KINDS, _norm_scope
+
+PROVIDERS = frozenset({"sentry", "glitchtip"})
+
+OUTCOME_PREVIEW = "preview"
+OUTCOME_LINKED = "linked_existing"
+OUTCOME_CREATED = "created_issue"
+OUTCOME_UPDATED = "updated_link"
+OUTCOME_BLOCKED = "blocked"
+OUTCOME_NO_ACTION = "no_safe_action"
+
+# Patterns scrubbed from any bridge-facing text (bodies, titles, tags, logs).
+_SECRET_PATTERNS: tuple[re.Pattern[str], ...] = (
+    re.compile(
+        r"(?i)\b(api[_-]?token|token|secret|password|passwd)\s*[:=]\s*\S+"
+    ),
+    re.compile(
+        r"(?i)\bAuthorization\s*[:=]\s*(?:Bearer\s+)?[A-Za-z0-9._\-+=/]{8,}"
+    ),
+    re.compile(r"(?i)\bBearer\s+[A-Za-z0-9._\-]{8,}"),
+    re.compile(r"(?i)\bBasic\s+[A-Za-z0-9+/=]{8,}"),
+    re.compile(r"(?i)\b(dsn|sentry_dsn|glitchtip_dsn)\s*[:=]\s*\S+"),
+    re.compile(r"https?://[^/\s]+:[^@/\s]+@"),  # user:pass@host
+    re.compile(r"(?i)\b(keychain[_-]?id|private[_-]?key)\s*[:=]\s*\S+"),
+    re.compile(r"(?i)cookie\s*[:=]\s*[^\s;]+"),
+    re.compile(r"(?i)\bsession[_-]?id\s*[:=]\s*\S+"),
+)
+
+_SENSITIVE_TAG_KEYS = frozenset(
+    {
+        "authorization",
+        "cookie",
+        "set-cookie",
+        "password",
+        "passwd",
+        "secret",
+        "token",
+        "api_key",
+        "apikey",
+        "dsn",
+        "sentry_dsn",
+        "access_token",
+        "refresh_token",
+    }
+)
+
+DEFAULT_LABELS = (
+    "type:bug",
+    "observability",
+    "status:ready",
+)
+
+
+@dataclass(frozen=True)
+class ProjectMapping:
+    """Maps a monitoring project to a Gitea repository."""
+
+    name: str
+    provider: str
+    monitor_base_url: str
+    monitor_org: str
+    monitor_project: str
+    gitea_org: str
+    gitea_repo: str
+    default_labels: tuple[str, ...] = DEFAULT_LABELS
+    environment_filters: tuple[str, ...] = ()
+    severity_threshold: str | None = None
+
+    def __post_init__(self) -> None:
+        prov = (self.provider or "").strip().lower()
+        if prov not in PROVIDERS:
+            raise ControlPlaneError(
+                f"unknown provider '{self.provider}'; expected one of {sorted(PROVIDERS)}"
+            )
+        object.__setattr__(self, "provider", prov)
+        object.__setattr__(self, "monitor_base_url", (self.monitor_base_url or "").rstrip("/"))
+        object.__setattr__(self, "monitor_org", (self.monitor_org or "").strip())
+        object.__setattr__(self, "monitor_project", (self.monitor_project or "").strip())
+        object.__setattr__(self, "gitea_org", (self.gitea_org or "").strip())
+        object.__setattr__(self, "gitea_repo", (self.gitea_repo or "").strip())
+        object.__setattr__(
+            self,
+            "default_labels",
+            tuple(str(x).strip() for x in (self.default_labels or DEFAULT_LABELS) if str(x).strip()),
+        )
+
+    def as_dict(self) -> dict[str, Any]:
+        return {
+            "name": self.name,
+            "provider": self.provider,
+            "monitor_base_url": self.monitor_base_url,
+            "monitor_org": self.monitor_org,
+            "monitor_project": self.monitor_project,
+            "gitea_org": self.gitea_org,
+            "gitea_repo": self.gitea_repo,
+            "default_labels": list(self.default_labels),
+            "environment_filters": list(self.environment_filters),
+            "severity_threshold": self.severity_threshold,
+        }
+
+    def matches_observation(self, obs: dict[str, Any]) -> bool:
+        if (obs.get("provider") or "").strip().lower() != self.provider:
+            return False
+        base = (obs.get("provider_base_url") or obs.get("monitor_base_url") or "").rstrip("/")
+        if base and self.monitor_base_url and base != self.monitor_base_url:
+            return False
+        org = (obs.get("provider_org") or obs.get("monitor_org") or "").strip()
+        if org and self.monitor_org and org != self.monitor_org:
+            return False
+        project = (obs.get("provider_project") or obs.get("monitor_project") or "").strip()
+        if project and self.monitor_project and project != self.monitor_project:
+            return False
+        return True
+
+
+@dataclass
+class NormalizedIncident:
+    provider: str
+    provider_base_url: str
+    provider_org: str
+    provider_project: str
+    provider_issue_id: str
+    provider_short_id: str | None = None
+    provider_permalink: str | None = None
+    fingerprint: str | None = None
+    first_seen: str | None = None
+    last_seen: str | None = None
+    event_count: int | None = None
+    status: str = "open"
+    environment: str | None = None
+    severity: str | None = None
+    culprit: str | None = None
+    title: str = ""
+    summary: str = ""
+    tags: dict[str, str] = field(default_factory=dict)
+    gitea_org: str = ""
+    gitea_repo: str = ""
+    default_labels: tuple[str, ...] = DEFAULT_LABELS
+
+    def provider_key(self) -> tuple[str, str, str, str, str]:
+        return (
+            self.provider,
+            _norm_scope(self.provider_base_url),
+            _norm_scope(self.provider_org),
+            _norm_scope(self.provider_project),
+            str(self.provider_issue_id),
+        )
+
+    def as_dict(self) -> dict[str, Any]:
+        return {
+            "provider": self.provider,
+            "provider_base_url": self.provider_base_url,
+            "provider_org": self.provider_org,
+            "provider_project": self.provider_project,
+            "provider_issue_id": self.provider_issue_id,
+            "provider_short_id": self.provider_short_id,
+            "provider_permalink": self.provider_permalink,
+            "fingerprint": self.fingerprint,
+            "first_seen": self.first_seen,
+            "last_seen": self.last_seen,
+            "event_count": self.event_count,
+            "status": self.status,
+            "environment": self.environment,
+            "severity": self.severity,
+            "culprit": self.culprit,
+            "title": self.title,
+            "summary": self.summary,
+            "tags": dict(self.tags),
+            "gitea_org": self.gitea_org,
+            "gitea_repo": self.gitea_repo,
+            "default_labels": list(self.default_labels),
+        }
+
+
+def redact_text(value: Any) -> str:
+    """Redact secret-like material from a string (fail closed to empty)."""
+    if value is None:
+        return ""
+    text = str(value)
+    for pat in _SECRET_PATTERNS:
+        text = pat.sub("[REDACTED]", text)
+    return text
+
+
+def sanitize_tags(tags: Any) -> dict[str, str]:
+    if not isinstance(tags, dict):
+        return {}
+    out: dict[str, str] = {}
+    for k, v in tags.items():
+        key = str(k).strip().lower()
+        if not key or key in _SENSITIVE_TAG_KEYS:
+            continue
+        if any(s in key for s in ("token", "secret", "password", "cookie", "auth", "dsn")):
+            continue
+        val = redact_text(v)
+        if "[REDACTED]" in val:
+            continue
+        if len(val) > 200:
+            val = val[:200] + "…"
+        out[key] = val
+    return out
+
+
+def project_mapping_from_dict(data: dict[str, Any]) -> ProjectMapping:
+    labels = data.get("default_labels") or DEFAULT_LABELS
+    envs = data.get("environment_filters") or ()
+    return ProjectMapping(
+        name=str(data.get("name") or data.get("monitor_project") or "unnamed"),
+        provider=str(data.get("provider") or ""),
+        monitor_base_url=str(data.get("monitor_base_url") or data.get("provider_base_url") or ""),
+        monitor_org=str(data.get("monitor_org") or data.get("provider_org") or ""),
+        monitor_project=str(data.get("monitor_project") or data.get("provider_project") or ""),
+        gitea_org=str(data.get("gitea_org") or ""),
+        gitea_repo=str(data.get("gitea_repo") or ""),
+        default_labels=tuple(labels),
+        environment_filters=tuple(envs),
+        severity_threshold=data.get("severity_threshold"),
+    )
+
+
+def load_project_mappings(
+    *,
+    config_path: str | None = None,
+    mappings_json: str | None = None,
+) -> list[ProjectMapping]:
+    """Load project mappings from JSON env, file, or explicit JSON string.
+
+    Env: ``GITEA_OBSERVABILITY_PROJECTS_JSON`` or path
+    ``GITEA_OBSERVABILITY_PROJECTS_FILE``.
+    """
+    raw: Any = None
+    if mappings_json:
+        raw = json.loads(mappings_json)
+    else:
+        env_json = (os.environ.get("GITEA_OBSERVABILITY_PROJECTS_JSON") or "").strip()
+        path = (
+            (config_path or "").strip()
+            or (os.environ.get("GITEA_OBSERVABILITY_PROJECTS_FILE") or "").strip()
+        )
+        if env_json:
+            raw = json.loads(env_json)
+        elif path and os.path.isfile(path):
+            with open(path, encoding="utf-8") as fh:
+                raw = json.load(fh)
+    if raw is None:
+        return []
+    if isinstance(raw, dict) and "projects" in raw:
+        raw = raw["projects"]
+    if not isinstance(raw, list):
+        raise ControlPlaneError("observability project mappings must be a list")
+    return [project_mapping_from_dict(item) for item in raw if isinstance(item, dict)]
+
+
+def resolve_mapping(
+    observation: dict[str, Any],
+    mappings: Sequence[ProjectMapping],
+    *,
+    explicit: ProjectMapping | None = None,
+) -> ProjectMapping | None:
+    if explicit is not None:
+        return explicit
+    matches = [m for m in mappings if m.matches_observation(observation)]
+    if len(matches) == 1:
+        return matches[0]
+    if len(matches) > 1:
+        raise ControlPlaneError(
+            "ambiguous project mapping for observation: "
+            + ", ".join(m.name for m in matches)
+            + " (fail closed, #612)"
+        )
+    # Fallback: observation itself may carry gitea targets
+    g_org = (observation.get("gitea_org") or "").strip()
+    g_repo = (observation.get("gitea_repo") or "").strip()
+    provider = (observation.get("provider") or "").strip().lower()
+    if g_org and g_repo and provider in PROVIDERS:
+        return ProjectMapping(
+            name=f"inline-{provider}",
+            provider=provider,
+            monitor_base_url=str(
+                observation.get("provider_base_url")
+                or observation.get("monitor_base_url")
+                or ""
+            ),
+            monitor_org=str(
+                observation.get("provider_org") or observation.get("monitor_org") or ""
+            ),
+            monitor_project=str(
+                observation.get("provider_project")
+                or observation.get("monitor_project")
+                or ""
+            ),
+            gitea_org=g_org,
+            gitea_repo=g_repo,
+            default_labels=tuple(observation.get("default_labels") or DEFAULT_LABELS),
+        )
+    return None
+
+
+def normalize_incident(
+    observation: dict[str, Any],
+    mapping: ProjectMapping,
+) -> NormalizedIncident:
+    """Normalize + sanitize a raw provider observation (fail closed)."""
+    if not isinstance(observation, dict):
+        raise ControlPlaneError("observation must be a dict (fail closed, #612)")
+
+    provider = (observation.get("provider") or mapping.provider or "").strip().lower()
+    if provider not in PROVIDERS:
+        raise ControlPlaneError(
+            f"unknown provider '{provider}'; expected {sorted(PROVIDERS)} (fail closed)"
+        )
+    if provider != mapping.provider:
+        raise ControlPlaneError(
+            f"observation provider '{provider}' does not match mapping "
+            f"'{mapping.provider}' (fail closed, #612)"
+        )
+
+    issue_id = observation.get("provider_issue_id") or observation.get("id")
+    if issue_id is None or str(issue_id).strip() == "":
+        raise ControlPlaneError(
+            "observation missing provider_issue_id (fail closed, #612)"
+        )
+
+    base = (
+        observation.get("provider_base_url")
+        or observation.get("monitor_base_url")
+        or mapping.monitor_base_url
+        or ""
+    )
+    org = (
+        observation.get("provider_org")
+        or observation.get("monitor_org")
+        or mapping.monitor_org
+        or ""
+    )
+    project = (
+        observation.get("provider_project")
+        or observation.get("monitor_project")
+        or mapping.monitor_project
+        or ""
+    )
+
+    title = redact_text(
+        observation.get("title")
+        or observation.get("culprit")
+        or f"{provider} incident {issue_id}"
+    )
+    meta = observation.get("metadata")
+    meta_value = meta.get("value") if isinstance(meta, dict) else None
+    raw_sum = (
+        observation.get("summary")
+        or observation.get("message")
+        or meta_value
+        or title
+    )
+    summary = redact_text(raw_sum)
+
+    permalink = observation.get("provider_permalink") or observation.get("permalink")
+    if permalink:
+        permalink = redact_text(permalink)
+        if "[REDACTED]" in permalink:
+            permalink = None
+
+    tags = sanitize_tags(observation.get("tags") or {})
+    event_count = observation.get("event_count") or observation.get("count")
+    try:
+        event_count_i = int(event_count) if event_count is not None else None
+    except (TypeError, ValueError):
+        event_count_i = None
+
+    return NormalizedIncident(
+        provider=provider,
+        provider_base_url=str(base).rstrip("/"),
+        provider_org=str(org).strip(),
+        provider_project=str(project).strip(),
+        provider_issue_id=str(issue_id).strip(),
+        provider_short_id=redact_text(observation.get("provider_short_id") or observation.get("shortId") or "")
+        or None,
+        provider_permalink=permalink,
+        fingerprint=redact_text(observation.get("fingerprint") or "") or None,
+        first_seen=str(observation.get("first_seen") or observation.get("firstSeen") or "")
+        or None,
+        last_seen=str(observation.get("last_seen") or observation.get("lastSeen") or "")
+        or None,
+        event_count=event_count_i,
+        status=str(observation.get("status") or "open").strip().lower() or "open",
+        environment=redact_text(observation.get("environment") or "") or None,
+        severity=redact_text(observation.get("severity") or observation.get("level") or "")
+        or None,
+        culprit=redact_text(observation.get("culprit") or "") or None,
+        title=title[:200],
+        summary=summary[:2000],
+        tags=tags,
+        gitea_org=mapping.gitea_org,
+        gitea_repo=mapping.gitea_repo,
+        default_labels=mapping.default_labels,
+    )
+
+
+def build_gitea_issue_title(inc: NormalizedIncident) -> str:
+    env = f" [{inc.environment}]" if inc.environment else ""
+    sev = f" ({inc.severity})" if inc.severity else ""
+    return redact_text(f"[obs:{inc.provider}]{env}{sev} {inc.title}")[:180]
+
+
+def build_gitea_issue_body(inc: NormalizedIncident) -> str:
+    """Sanitized durable issue body (no secrets)."""
+    lines = [
+        "## Observability incident (bridge #612)",
+        "",
+        "<!-- mcp-incident-bridge:v1 -->",
+        f"<!-- provider={inc.provider} issue_id={inc.provider_issue_id} -->",
+        "",
+        f"- **provider:** `{inc.provider}`",
+        f"- **provider_issue_id:** `{inc.provider_issue_id}`",
+    ]
+    if inc.provider_short_id:
+        lines.append(f"- **provider_short_id:** `{inc.provider_short_id}`")
+    if inc.provider_permalink:
+        lines.append(f"- **provider_url:** {inc.provider_permalink}")
+    lines.extend(
+        [
+            f"- **provider_org/project:** `{inc.provider_org}` / `{inc.provider_project}`",
+            f"- **base_url:** `{inc.provider_base_url}`",
+            f"- **fingerprint:** `{inc.fingerprint or ''}`",
+            f"- **first_seen:** `{inc.first_seen or ''}`",
+            f"- **last_seen:** `{inc.last_seen or ''}`",
+            f"- **event_count:** `{inc.event_count if inc.event_count is not None else ''}`",
+            f"- **environment:** `{inc.environment or ''}`",
+            f"- **severity:** `{inc.severity or ''}`",
+            f"- **culprit:** `{inc.culprit or ''}`",
+            f"- **status:** `{inc.status}`",
+            "",
+            "### Summary",
+            "",
+            redact_text(inc.summary) or "(no summary)",
+            "",
+            "### Safe tags",
+            "",
+        ]
+    )
+    if inc.tags:
+        for k, v in sorted(inc.tags.items()):
+            lines.append(f"- `{k}`: `{v}`")
+    else:
+        lines.append("- (none)")
+    lines.extend(
+        [
+            "",
+            "### Canonical next action",
+            "",
+            "Author: investigate and fix under normal Gitea workflow. "
+            "Allocator may select this issue as ordinary Gitea work "
+            "(never as a raw provider incident).",
+            "",
+            "### Bridge notes",
+            "",
+            "- Raw Sentry/GlitchTip incidents are **not** assignable work items.",
+            "- Gitea remains the durable work record; DB stores `incident_links` only.",
+            "- Provider tokens must never appear in this issue.",
+        ]
+    )
+    return "\n".join(lines)
+
+
+def _link_conflict(existing: dict[str, Any], inc: NormalizedIncident) -> str | None:
+    """Fail closed if existing link targets a different Gitea issue/repo."""
+    eg_org = str(existing.get("gitea_org") or "")
+    eg_repo = str(existing.get("gitea_repo") or "")
+    eg_num = existing.get("gitea_issue_number")
+    if eg_org and eg_org != inc.gitea_org:
+        return (
+            f"existing link points to org '{eg_org}', mapping wants "
+            f"'{inc.gitea_org}' (fail closed, #612)"
+        )
+    if eg_repo and eg_repo != inc.gitea_repo:
+        return (
+            f"existing link points to repo '{eg_repo}', mapping wants "
+            f"'{inc.gitea_repo}' (fail closed, #612)"
+        )
+    # fingerprint conflict when both present and disagree
+    ef = (existing.get("fingerprint") or "").strip()
+    nf = (inc.fingerprint or "").strip()
+    if ef and nf and ef != nf:
+        # Same provider issue id with different fingerprints is ambiguous.
+        return (
+            f"fingerprint conflict for provider issue {inc.provider_issue_id}: "
+            f"link has '{ef}', observation has '{nf}' (fail closed, #612)"
+        )
+    return None
+
+
+CreateIssueFn = Callable[[str, str, list[str], str, str], dict[str, Any]]
+# create_issue_fn(title, body, labels, gitea_org, gitea_repo) -> {"number": int, ...}
+
+
+def reconcile_incident(
+    db: ControlPlaneDB | None,
+    *,
+    observation: dict[str, Any],
+    mappings: Sequence[ProjectMapping] | None = None,
+    mapping: ProjectMapping | None = None,
+    apply: bool = False,
+    create_issue_fn: CreateIssueFn | None = None,
+    force_gitea_issue_number: int | None = None,
+) -> dict[str, Any]:
+    """Reconcile one observation into incident_links + optional Gitea issue.
+
+    *apply=False* (default): preview only — no DB mutation, no Gitea mutation.
+    *apply=True*: upsert link; create Gitea issue when none linked (requires
+    ``create_issue_fn``) or use ``force_gitea_issue_number`` for explicit link.
+
+    Never creates control-plane ``work_items`` for raw incidents.
+    """
+    base: dict[str, Any] = {
+        "success": False,
+        "apply": bool(apply),
+        "outcome": OUTCOME_NO_ACTION,
+        "performed": False,
+        "gitea_mutated": False,
+        "db_mutated": False,
+        "raw_incident_assignable": False,
+        "work_item_kind_would_be": None,
+        "reasons": [],
+        "skipped": None,
+        "incident": None,
+        "existing_link": None,
+        "gitea_issue": None,
+        "action": None,
+        "mapping": None,
+        "substrate": "control_plane_db.incident_links",
+        "durable_work_system": "gitea_issues",
+    }
+
+    if db is None:
+        base["reasons"] = [
+            "control-plane DB substrate unavailable (fail closed, #613/#612)"
+        ]
+        base["outcome"] = OUTCOME_BLOCKED
+        return base
+
+    try:
+        maps = list(mappings or [])
+        resolved = resolve_mapping(observation, maps, explicit=mapping)
+        if resolved is None:
+            base["outcome"] = OUTCOME_BLOCKED
+            base["reasons"] = [
+                "no project mapping for observation (fail closed, #612); "
+                "configure GITEA_OBSERVABILITY_PROJECTS_JSON or pass mapping"
+            ]
+            return base
+        base["mapping"] = resolved.as_dict()
+        inc = normalize_incident(observation, resolved)
+        base["incident"] = inc.as_dict()
+    except (ControlPlaneError, json.JSONDecodeError, TypeError, ValueError) as exc:
+        base["outcome"] = OUTCOME_BLOCKED
+        base["reasons"] = [str(exc)]
+        return base
+
+    # Environment filter (optional)
+    if resolved.environment_filters and inc.environment:
+        allowed = {e.lower() for e in resolved.environment_filters}
+        if inc.environment.lower() not in allowed:
+            base["outcome"] = OUTCOME_NO_ACTION
+            base["reasons"] = [
+                f"environment '{inc.environment}' not in filters "
+                f"{sorted(allowed)}; skip (policy)"
+            ]
+            base["success"] = True
+            base["action"] = "skip_environment_filter"
+            return base
+
+    existing = db.get_incident_link_by_provider(
+        provider=inc.provider,
+        provider_issue_id=inc.provider_issue_id,
+        provider_base_url=inc.provider_base_url,
+        provider_org=inc.provider_org,
+        provider_project=inc.provider_project,
+    )
+    base["existing_link"] = existing
+
+    if existing:
+        conflict = _link_conflict(existing, inc)
+        if conflict:
+            base["outcome"] = OUTCOME_BLOCKED
+            base["reasons"] = [conflict]
+            return base
+
+    title = build_gitea_issue_title(inc)
+    body = build_gitea_issue_body(inc)
+    labels = list(inc.default_labels)
+    if inc.provider and f"{inc.provider}" not in labels:
+        labels.append(inc.provider)
+    if "observability" not in labels:
+        labels.append("observability")
+
+    preview_issue = {
+        "title": title,
+        "body_preview": body[:500] + ("…" if len(body) > 500 else ""),
+        "labels": labels,
+        "gitea_org": inc.gitea_org,
+        "gitea_repo": inc.gitea_repo,
+        "would_create": existing is None and force_gitea_issue_number is None,
+        "would_reuse_issue": int(existing["gitea_issue_number"])
+        if existing
+        else force_gitea_issue_number,
+    }
+    base["gitea_issue_preview"] = preview_issue
+
+    if not apply:
+        base["success"] = True
+        base["outcome"] = OUTCOME_PREVIEW
+        base["action"] = (
+            "preview_reuse_link" if existing else "preview_create_issue"
+        )
+        base["reasons"] = [
+            "dry-run only (apply=false); no Gitea mutation and no DB write — "
+            "call again with apply=true to create/link"
+        ]
+        if existing:
+            base["gitea_issue"] = {
+                "number": existing.get("gitea_issue_number"),
+                "org": existing.get("gitea_org"),
+                "repo": existing.get("gitea_repo"),
+            }
+        # Prove we never treat this as work_item kind incident
+        base["raw_incident_assignable"] = False
+        base["work_item_kind_would_be"] = "issue"  # only after Gitea issue exists
+        return base
+
+    # --- apply path ---
+    issue_number: int | None = None
+    created = False
+    if existing:
+        issue_number = int(existing["gitea_issue_number"])
+        action = "updated_existing_link"
+        outcome = OUTCOME_UPDATED
+    elif force_gitea_issue_number is not None:
+        issue_number = int(force_gitea_issue_number)
+        action = "link_explicit_issue"
+        outcome = OUTCOME_LINKED
+    else:
+        if create_issue_fn is None:
+            base["outcome"] = OUTCOME_BLOCKED
+            base["reasons"] = [
+                "apply=true requires create_issue_fn when no existing link "
+                "(fail closed, #612)"
+            ]
+            return base
+        try:
+            created_res = create_issue_fn(
+                title, body, labels, inc.gitea_org, inc.gitea_repo
+            )
+        except Exception as exc:  # noqa: BLE001
+            base["outcome"] = OUTCOME_BLOCKED
+            base["reasons"] = [
+                f"Gitea issue creation failed: {redact_text(exc)} (fail closed)"
+            ]
+            return base
+        number = created_res.get("number") if isinstance(created_res, dict) else None
+        if number is None:
+            base["outcome"] = OUTCOME_BLOCKED
+            base["reasons"] = [
+                "Gitea issue creation returned no issue number (fail closed, #612)"
+            ]
+            base["create_result"] = created_res
+            return base
+        issue_number = int(number)
+        created = True
+        action = "created_gitea_issue"
+        outcome = OUTCOME_CREATED
+        base["gitea_mutated"] = True
+        base["create_result"] = {
+            k: created_res.get(k)
+            for k in ("number", "success", "performed", "reasons")
+            if isinstance(created_res, dict)
+        }
+
+    assert issue_number is not None
+    try:
+        link = db.upsert_incident_link(
+            provider=inc.provider,
+            provider_issue_id=inc.provider_issue_id,
+            gitea_org=inc.gitea_org,
+            gitea_repo=inc.gitea_repo,
+            gitea_issue_number=issue_number,
+            provider_base_url=inc.provider_base_url,
+            provider_org=inc.provider_org,
+            provider_project=inc.provider_project,
+            provider_short_id=inc.provider_short_id,
+            provider_permalink=inc.provider_permalink,
+            fingerprint=inc.fingerprint,
+            first_seen=inc.first_seen,
+            last_seen=inc.last_seen,
+            event_count=inc.event_count,
+            status=inc.status if not created else "open",
+        )
+    except Exception as exc:  # noqa: BLE001
+        base["outcome"] = OUTCOME_BLOCKED
+        base["reasons"] = [
+            f"incident_links upsert failed: {redact_text(exc)} (fail closed, #613)"
+        ]
+        base["gitea_issue"] = {
+            "number": issue_number,
+            "org": inc.gitea_org,
+            "repo": inc.gitea_repo,
+            "created": created,
+        }
+        return base
+
+    base["success"] = True
+    base["performed"] = True
+    base["db_mutated"] = True
+    base["outcome"] = outcome
+    base["action"] = action
+    base["gitea_issue"] = {
+        "number": issue_number,
+        "org": inc.gitea_org,
+        "repo": inc.gitea_repo,
+        "created": created,
+    }
+    base["link"] = {
+        k: link.get(k)
+        for k in (
+            "link_id",
+            "provider",
+            "provider_issue_id",
+            "gitea_org",
+            "gitea_repo",
+            "gitea_issue_number",
+            "fingerprint",
+            "event_count",
+            "status",
+            "last_sync_at",
+        )
+    }
+    base["reasons"] = [
+        (
+            f"reused Gitea issue #{issue_number} for provider "
+            f"{inc.provider}:{inc.provider_issue_id}"
+            if not created
+            else f"created Gitea issue #{issue_number} and stored incident_links row"
+        ),
+        "raw provider incident is not an assignable work_item "
+        f"(WORK_KINDS={sorted(WORK_KINDS)})",
+    ]
+    base["raw_incident_assignable"] = False
+    base["work_item_kind_would_be"] = "issue"
+    base["allocator_visible_as"] = {
+        "kind": "issue",
+        "number": issue_number,
+        "org": inc.gitea_org,
+        "repo": inc.gitea_repo,
+        "note": "allocator selects normal Gitea issues only (#600/#612)",
+    }
+    return base
+
+
+def assert_not_raw_incident_work_item(kind: str) -> None:
+    """Guard used by tests / callers: incidents must not enter WORK_KINDS."""
+    k = (kind or "").strip().lower()
+    if k not in WORK_KINDS:
+        if k in ("sentry_incident", "glitchtip_incident", "incident", "observation"):
+            raise ControlPlaneError(
+                f"kind '{k}' is not assignable; bridge must create Gitea issues first (#612)"
+            )
+        raise ControlPlaneError(f"kind '{k}' is not assignable")

issue_content_gate.py

@@ -0,0 +1,180 @@
+"""Pre-create issue content gate (#582).
+
+Blocks issue creation when the title/body is a *vague reference* to
+out-of-band content the LLM cannot prove it has ("the drafted issue",
+"the prepared issue", "the issue we discussed", "the previous draft")
+with no durable source pointer.
+
+The rule from #582: an LLM must not fabricate an issue from stale chat
+memory. When the requested issue content is a vague reference and no
+durable source pointer (existing issue/PR/comment, checked-in file, URL,
+or scratchpad path) is present, the gate fails closed and the caller must
+return BLOCKED + DIAGNOSE naming the exact vague/missing fields.
+
+This gate intentionally does NOT require a non-empty body: title-only
+issue creation remains allowed for callers that explicitly want it. It
+only blocks content that is a placeholder reference to something the
+current context does not contain.
+"""
+
+from __future__ import annotations
+
+import re
+import unicodedata
+
+# Vague reference phrases that point at out-of-band draft content. Stored
+# normalized (lowercase, punctuation-stripped, whitespace-collapsed) so they
+# can be matched against normalized field text.
+VAGUE_REFERENCE_PHRASES: tuple[str, ...] = (
+    "the drafted issue",
+    "drafted issue",
+    "the prepared issue",
+    "prepared issue",
+    "the issue we discussed",
+    "the issue we talked about",
+    "the one we discussed",
+    "the previous draft",
+    "previous draft",
+    "the draft above",
+    "the issue above",
+    "as discussed",
+    "as we discussed",
+    "as previously discussed",
+    "per our discussion",
+    "per our conversation",
+    "per our chat",
+    "see above",
+    "same as before",
+    "the issue from earlier",
+    "the issue i drafted",
+    "the issue you drafted",
+)
+
+# A field that only restates a vague phrase (optionally with a leading verb
+# such as "create"/"add"/"open"/"file") is still a vague reference.
+_LEADING_VERBS = ("create", "add", "open", "file", "make", "please", "the")
+
+# Durable source pointers: if any of these appears, the content points at a
+# retrievable source of truth and is NOT treated as a fabricated reference.
+_DURABLE_SOURCE_REGEXES: tuple[re.Pattern[str], ...] = (
+    re.compile(r"#\d+"),                                   # #402
+    re.compile(r"\b(?:issue|pull|pr)\s+#?\d+", re.I),      # issue 402 / PR #17
+    re.compile(r"\bcomment\s+#?\d+", re.I),                # comment 8155
+    re.compile(r"https?://\S+", re.I),                     # URL
+    re.compile(                                            # checked-in file
+        r"[\w./-]+\.(?:py|md|json|txt|sh|ya?ml|toml|cfg|ini|rst)\b", re.I
+    ),
+    re.compile(r"\bscratchpad\b", re.I),                   # scratchpad path
+    re.compile(r"(?:^|\s)/[\w./-]+"),                      # absolute path
+)
+
+
+def normalize_text(text: str) -> str:
+    """Lowercase, punctuation-stripped, whitespace-collapsed text."""
+    norm = unicodedata.normalize("NFKC", (text or "").strip().lower())
+    norm = re.sub(r"[^\w\s]", " ", norm)
+    norm = re.sub(r"\s+", " ", norm).strip()
+    return norm
+
+
+def has_durable_source_pointer(text: str) -> bool:
+    """True when the raw text references a durable, retrievable source."""
+    raw = text or ""
+    return any(rx.search(raw) for rx in _DURABLE_SOURCE_REGEXES)
+
+
+def find_vague_reference(text: str) -> str | None:
+    """Return the vague phrase a field is dominated by, else ``None``.
+
+    A field is a vague reference only when it contains a known vague phrase,
+    carries no durable source pointer, and the phrase dominates the field
+    (the field is essentially just the phrase). Long, specific bodies that
+    merely contain "as discussed" in passing are not blocked.
+    """
+    if has_durable_source_pointer(text):
+        return None
+    norm = normalize_text(text)
+    if not norm:
+        return None
+    stripped = norm
+    for verb in _LEADING_VERBS:
+        if stripped.startswith(verb + " "):
+            stripped = stripped[len(verb) + 1:]
+    for phrase in VAGUE_REFERENCE_PHRASES:
+        if phrase in norm and len(stripped) <= len(phrase) + 12:
+            return phrase
+    return None
+
+
+def assess_issue_content(
+    title: str,
+    body: str = "",
+    *,
+    allow_incomplete: bool = False,
+) -> dict:
+    """Evaluate whether proposed issue content is durable enough to create.
+
+    Returns a dict with ``complete`` (bool), ``missing_fields``,
+    ``vague_fields``, ``reasons``, ``diagnose`` (joined reasons), and
+    ``has_durable_source_pointer``. ``allow_incomplete`` is an explicit
+    operator override that forces ``complete`` True.
+    """
+    t = (title or "").strip()
+    b = (body or "").strip()
+    missing_fields: list[str] = []
+    vague_fields: list[str] = []
+    reasons: list[str] = []
+
+    if not t:
+        missing_fields.append("title")
+        reasons.append("issue title is required")
+    else:
+        phrase = find_vague_reference(t)
+        if phrase:
+            vague_fields.append("title")
+            reasons.append(
+                f"title is a vague reference ('{phrase}') with no durable "
+                "content or source pointer"
+            )
+
+    if b:
+        phrase = find_vague_reference(b)
+        if phrase:
+            vague_fields.append("body")
+            reasons.append(
+                f"body is a vague reference ('{phrase}') with no durable "
+                "content or source pointer"
+            )
+
+    complete = allow_incomplete or (not missing_fields and not vague_fields)
+    return {
+        "complete": complete,
+        "missing_fields": missing_fields,
+        "vague_fields": vague_fields,
+        "reasons": reasons,
+        "diagnose": "; ".join(reasons),
+        "has_durable_source_pointer": has_durable_source_pointer(b)
+        or has_durable_source_pointer(t),
+        "override_applied": bool(
+            allow_incomplete and (missing_fields or vague_fields)
+        ),
+    }
+
+
+def pre_create_issue_content_gate(
+    title: str,
+    body: str = "",
+    *,
+    allow_incomplete: bool = False,
+) -> dict:
+    """Content gate wrapper mirroring the duplicate gate shape.
+
+    ``performed`` is True when the content is durable enough to create (or an
+    explicit override was supplied). When False, the caller must return
+    BLOCKED + DIAGNOSE and must not create the issue.
+    """
+    assessment = assess_issue_content(
+        title, body, allow_incomplete=allow_incomplete
+    )
+    assessment["performed"] = assessment["complete"]
+    return assessment

issue_lock_store.py

@@ -375,6 +375,64 @@ def _same_realpath(left: str | None, right: str | None) -> bool:
         return left == right
 
 
+
+def assess_expired_lock_reclaim(
+    existing_lock: dict[str, Any] | None,
+    *,
+    now: datetime | None = None,
+) -> dict[str, Any]:
+    """Decide whether an expired/stale author issue lock may be reclaimed (#601).
+
+    Required proof (any reclaim of non-live lock):
+    * lease not live (expired or dead pid)
+    * owner process dead OR worktree missing
+    * no force-delete of live foreign ownership
+    """
+    if not existing_lock:
+        return {
+            "reclaim_allowed": True,
+            "reasons": ["no existing lock"],
+            "freshness": assess_lock_freshness(None, now=now),
+        }
+    freshness = assess_lock_freshness(existing_lock, now=now)
+    if freshness.get("live"):
+        return {
+            "reclaim_allowed": False,
+            "reasons": ["lock is still live; cannot reclaim (fail closed)"],
+            "freshness": freshness,
+        }
+    pid = existing_lock.get("session_pid")
+    if pid is None:
+        pid = existing_lock.get("pid")
+    dead = not is_process_alive(pid) if pid is not None else True
+    wt = str(existing_lock.get("worktree_path") or "")
+    missing_wt = (not wt) or (not os.path.isdir(os.path.realpath(wt)))
+    if not (dead or missing_wt):
+        return {
+            "reclaim_allowed": False,
+            "reasons": [
+                "expired/stale lock still has live owner pid and present worktree; "
+                "recovery review required (fail closed)"
+            ],
+            "freshness": freshness,
+            "owner_pid_dead": dead,
+            "worktree_missing": missing_wt,
+        }
+    return {
+        "reclaim_allowed": True,
+        "reasons": [
+            "non-live lock with dead process and/or missing worktree; "
+            "sanctioned reclaim allowed"
+        ],
+        "freshness": freshness,
+        "owner_pid_dead": dead,
+        "worktree_missing": missing_wt,
+        "prior_branch": existing_lock.get("branch_name"),
+        "prior_worktree": existing_lock.get("worktree_path"),
+        "prior_pid": pid,
+    }
+
+
 def assess_same_issue_lease_conflict(
     existing_lock: dict[str, Any] | None,
     *,
@@ -405,6 +463,11 @@ def assess_same_issue_lease_conflict(
         and _same_realpath(str(existing_worktree or ""), worktree_path)
     )
     if is_lease_expired(existing_lock, now=now):
+        reclaim = assess_expired_lock_reclaim(existing_lock, now=now)
+        if reclaim.get("reclaim_allowed"):
+            # #601: expired + dead pid / missing worktree may be reclaimed
+            # through the normal lock path (sanctioned overwrite).
+            return None
         return (
             f"Issue #{issue_number} has an expired {operation_type} lease on "
             f"branch '{existing_branch}' from worktree '{existing_worktree}'. "

issue_workflow_labels.py

@@ -0,0 +1,406 @@
+"""Canonical issue type/status label taxonomy and transition helpers (#513)."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import Iterable, Mapping, Sequence
+
+
+@dataclass(frozen=True)
+class LabelSpec:
+    name: str
+    color: str
+    description: str
+
+
+TYPE_LABEL_SPECS: tuple[LabelSpec, ...] = (
+    LabelSpec("type:bug", "b60205", "Bug or defect"),
+    LabelSpec("type:feature", "a2eeef", "Feature or enhancement"),
+    LabelSpec("type:process", "5319e7", "Process or policy work"),
+    LabelSpec("type:workflow", "c2e0c6", "Workflow automation or guidance"),
+    LabelSpec("type:guardrail", "d93f0b", "Safety gate or guardrail"),
+    LabelSpec("type:docs", "006b75", "Documentation work"),
+    LabelSpec("type:test", "0e8a16", "Tests or test infrastructure"),
+    LabelSpec("type:discussion", "bfdadc", "Discussion-only issue"),
+    LabelSpec("type:umbrella", "c5def5", "Umbrella or tracker issue"),
+    LabelSpec("type:cleanup", "ededed", "Cleanup or hygiene work"),
+)
+
+STATUS_LABEL_SPECS: tuple[LabelSpec, ...] = (
+    LabelSpec("status:triage", "fbca04", "Issue needs triage"),
+    LabelSpec("status:ready", "0e8a16", "Issue is ready for work"),
+    LabelSpec("status:claimed", "fefe2e", "Issue is claimed"),
+    LabelSpec("status:in-progress", "fefe2e", "Issue is being worked on"),
+    LabelSpec("status:blocked", "b60205", "Issue is blocked"),
+    LabelSpec("status:needs-review", "0052cc", "Issue work needs review"),
+    LabelSpec("status:pr-open", "1d76db", "A linked PR is open"),
+    LabelSpec(
+        "status:changes-requested",
+        "e11d21",
+        "Reviewer requested changes on the linked PR",
+    ),
+    LabelSpec("status:approved", "0e8a16", "Linked PR is approved"),
+    LabelSpec("status:merged", "5319e7", "Linked PR is merged"),
+    LabelSpec("status:reconcile", "d93f0b", "Issue needs reconciliation"),
+    LabelSpec("status:done", "0e8a16", "Issue workflow is complete"),
+    LabelSpec("status:duplicate", "cccccc", "Issue is a duplicate"),
+    LabelSpec("status:wontfix", "000000", "Issue will not be fixed"),
+)
+
+# Durable validation-outcome labels (#529): the four canonical distinctions a
+# reviewer report can carry. These are orthogonal to the single active status:*
+# label, so they use their own prefix and are not subject to the one-status
+# invariant.
+VALIDATION_LABEL_SPECS: tuple[LabelSpec, ...] = (
+    LabelSpec("validation:clean-pass", "0e8a16", "Validation was a clean pass"),
+    LabelSpec(
+        "validation:baseline-accepted",
+        "fbca04",
+        "Validation passed with a baseline-proven unrelated failure",
+    ),
+    LabelSpec(
+        "validation:blocked",
+        "b60205",
+        "Validation blocked by an unresolved failure",
+    ),
+    LabelSpec(
+        "validation:post-merge-moot",
+        "5319e7",
+        "Validation is post-merge moot (PR already merged/closed before review)",
+    ),
+)
+
+# Lifecycle role-ownership labels (#603): which workflow role currently owns the
+# item. Advisory visibility only — the control-plane lease (#601) remains the
+# source of truth for mutation authority. Only one role:* label is active at a
+# time, mirroring the single-active-status invariant.
+ROLE_LABEL_SPECS: tuple[LabelSpec, ...] = (
+    LabelSpec("role:author", "1d76db", "Author currently owns the item"),
+    LabelSpec("role:reviewer", "5319e7", "Reviewer currently owns the item"),
+    LabelSpec("role:merger", "0e8a16", "Merger currently owns the item"),
+)
+
+# Lifecycle hazard labels (#603): orthogonal warning flags surfacing dangerous
+# coordination conditions. Unlike status/role, multiple hazard:* labels may be
+# active at once, and they never substitute for live lease / PR state checks.
+HAZARD_LABEL_SPECS: tuple[LabelSpec, ...] = (
+    LabelSpec("hazard:stale-lease", "d93f0b", "A stale or expired lease references this item"),
+    LabelSpec(
+        "hazard:workflow-contaminated",
+        "b60205",
+        "Session/workflow state is contaminated and must not mutate",
+    ),
+    LabelSpec("hazard:conflicted", "e11d21", "Linked PR has merge conflicts"),
+    LabelSpec("hazard:root-mutation", "b60205", "Work was mutated in the project root checkout"),
+    LabelSpec("hazard:manual-state", "d93f0b", "Session or lease state was edited manually"),
+    LabelSpec(
+        "hazard:terminal-blocker",
+        "000000",
+        "A terminal review/merge lock blocks progress (#332/#602)",
+    ),
+)
+
+CANONICAL_LABEL_SPECS: tuple[LabelSpec, ...] = (
+    TYPE_LABEL_SPECS
+    + STATUS_LABEL_SPECS
+    + VALIDATION_LABEL_SPECS
+    + ROLE_LABEL_SPECS
+    + HAZARD_LABEL_SPECS
+)
+
+TYPE_LABELS: frozenset[str] = frozenset(spec.name for spec in TYPE_LABEL_SPECS)
+STATUS_LABELS: frozenset[str] = frozenset(spec.name for spec in STATUS_LABEL_SPECS)
+VALIDATION_LABELS: frozenset[str] = frozenset(
+    spec.name for spec in VALIDATION_LABEL_SPECS
+)
+ROLE_LABELS: frozenset[str] = frozenset(spec.name for spec in ROLE_LABEL_SPECS)
+HAZARD_LABELS: frozenset[str] = frozenset(spec.name for spec in HAZARD_LABEL_SPECS)
+CANONICAL_LABELS: frozenset[str] = frozenset(
+    spec.name for spec in CANONICAL_LABEL_SPECS
+)
+
+STATUS_TRANSITIONS: dict[str, str] = {
+    "triage": "status:triage",
+    "ready": "status:ready",
+    "claim": "status:claimed",
+    "claimed": "status:claimed",
+    "start": "status:in-progress",
+    "start_work": "status:in-progress",
+    "in_progress": "status:in-progress",
+    "in-progress": "status:in-progress",
+    "block": "status:blocked",
+    "blocked": "status:blocked",
+    "needs_review": "status:needs-review",
+    "needs-review": "status:needs-review",
+    "reviewing": "status:needs-review",
+    "pr_open": "status:pr-open",
+    "pr-open": "status:pr-open",
+    "changes_requested": "status:changes-requested",
+    "changes-requested": "status:changes-requested",
+    "changes": "status:changes-requested",
+    "approved": "status:approved",
+    "merge_ready": "status:approved",
+    "merge-ready": "status:approved",
+    "merge": "status:reconcile",
+    "merged": "status:reconcile",
+    "reconcile": "status:reconcile",
+    "done": "status:done",
+    "complete": "status:done",
+    "duplicate": "status:duplicate",
+    "wontfix": "status:wontfix",
+    "abandoned": "status:wontfix",
+    # #603 requested state:* synonyms folded into the canonical status vocabulary
+    "needs_triage": "status:triage",
+    "needs-triage": "status:triage",
+    "authoring": "status:in-progress",
+}
+
+# #603: single-active role ownership. Maps role kinds / role:* labels to the
+# canonical role label. Mirrors STATUS_TRANSITIONS for the role dimension.
+ROLE_TRANSITIONS: dict[str, str] = {
+    "author": "role:author",
+    "reviewer": "role:reviewer",
+    "merger": "role:merger",
+}
+
+# #603: hazard flag synonyms. Hazards are additive (not single-active), so this
+# only normalizes names; it does not drive replacement.
+HAZARD_TRANSITIONS: dict[str, str] = {
+    "stale_lease": "hazard:stale-lease",
+    "stale-lease": "hazard:stale-lease",
+    "workflow_contaminated": "hazard:workflow-contaminated",
+    "workflow-contaminated": "hazard:workflow-contaminated",
+    "contaminated": "hazard:workflow-contaminated",
+    "conflicted": "hazard:conflicted",
+    "conflict": "hazard:conflicted",
+    "root_mutation": "hazard:root-mutation",
+    "root-mutation": "hazard:root-mutation",
+    "manual_state": "hazard:manual-state",
+    "manual-state": "hazard:manual-state",
+    "terminal_blocker": "hazard:terminal-blocker",
+    "terminal-blocker": "hazard:terminal-blocker",
+}
+
+
+def label_name(label: str | Mapping[str, object]) -> str:
+    """Return a normalized label name from a Gitea label object or string."""
+    if isinstance(label, str):
+        return label.strip()
+    value = label.get("name")
+    return str(value or "").strip()
+
+
+def label_names(labels: Iterable[str | Mapping[str, object]] | Mapping[str, object]) -> list[str]:
+    """Extract label names from a label list or issue-like mapping."""
+    raw: object
+    if isinstance(labels, Mapping):
+        raw = labels.get("labels", [])
+    else:
+        raw = labels
+    return [name for name in (label_name(label) for label in raw or []) if name]
+
+
+def type_labels(labels: Iterable[str | Mapping[str, object]] | Mapping[str, object]) -> list[str]:
+    return [name for name in label_names(labels) if name.startswith("type:")]
+
+
+def status_labels(labels: Iterable[str | Mapping[str, object]] | Mapping[str, object]) -> list[str]:
+    return [name for name in label_names(labels) if name.startswith("status:")]
+
+
+def canonical_status_label(status_or_transition: str) -> str:
+    status = status_or_transition.strip()
+    if status in STATUS_LABELS:
+        return status
+    normalized = status.lower().replace(" ", "-")
+    try:
+        return STATUS_TRANSITIONS[normalized]
+    except KeyError as exc:
+        raise ValueError(
+            f"unknown workflow status or transition '{status_or_transition}'"
+        ) from exc
+
+
+def transition_status_labels(
+    existing_labels: Iterable[str | Mapping[str, object]] | Mapping[str, object],
+    status_or_transition: str,
+) -> list[str]:
+    """Replace all active status labels with the requested canonical status."""
+    new_status = canonical_status_label(status_or_transition)
+    kept = [name for name in label_names(existing_labels) if not name.startswith("status:")]
+    if new_status not in kept:
+        kept.append(new_status)
+    return kept
+
+
+def role_labels(labels: Iterable[str | Mapping[str, object]] | Mapping[str, object]) -> list[str]:
+    return [name for name in label_names(labels) if name.startswith("role:")]
+
+
+def hazard_labels(labels: Iterable[str | Mapping[str, object]] | Mapping[str, object]) -> list[str]:
+    return [name for name in label_names(labels) if name.startswith("hazard:")]
+
+
+def canonical_role_label(role_or_transition: str) -> str:
+    role = role_or_transition.strip()
+    if role in ROLE_LABELS:
+        return role
+    normalized = role.lower().replace(" ", "-")
+    try:
+        return ROLE_TRANSITIONS[normalized]
+    except KeyError as exc:
+        raise ValueError(
+            f"unknown workflow role or transition '{role_or_transition}'"
+        ) from exc
+
+
+def transition_role_labels(
+    existing_labels: Iterable[str | Mapping[str, object]] | Mapping[str, object],
+    role_or_transition: str,
+) -> list[str]:
+    """Replace all active role labels with the requested canonical role."""
+    new_role = canonical_role_label(role_or_transition)
+    kept = [name for name in label_names(existing_labels) if not name.startswith("role:")]
+    if new_role not in kept:
+        kept.append(new_role)
+    return kept
+
+
+def canonical_hazard_label(hazard: str) -> str:
+    name = hazard.strip()
+    if name in HAZARD_LABELS:
+        return name
+    normalized = name.lower().replace(" ", "-")
+    try:
+        return HAZARD_TRANSITIONS[normalized]
+    except KeyError as exc:
+        raise ValueError(f"unknown workflow hazard '{hazard}'") from exc
+
+
+def add_hazard_label(
+    existing_labels: Iterable[str | Mapping[str, object]] | Mapping[str, object],
+    hazard: str,
+) -> list[str]:
+    """Add a hazard flag without disturbing status/role/type labels (additive)."""
+    new_hazard = canonical_hazard_label(hazard)
+    kept = label_names(existing_labels)
+    if new_hazard not in kept:
+        kept.append(new_hazard)
+    return kept
+
+
+def clear_hazard_label(
+    existing_labels: Iterable[str | Mapping[str, object]] | Mapping[str, object],
+    hazard: str,
+) -> list[str]:
+    """Remove a single hazard flag, leaving all other labels intact."""
+    target = canonical_hazard_label(hazard)
+    return [name for name in label_names(existing_labels) if name != target]
+
+
+def is_discussion(labels: Iterable[str | Mapping[str, object]] | Mapping[str, object]) -> bool:
+    return "type:discussion" in type_labels(labels)
+
+
+def is_implementation_candidate(
+    labels: Iterable[str | Mapping[str, object]] | Mapping[str, object],
+) -> bool:
+    """Whether an item may enter an implementation queue on labels alone.
+
+    Discussion issues are excluded (#603 AC3) unless a controller explicitly
+    selects them; the allocator still cross-checks live lease/PR state and never
+    trusts labels alone (#603 AC2).
+    """
+    return not is_discussion(labels)
+
+
+def requires_blocking_reason(
+    labels: Iterable[str | Mapping[str, object]] | Mapping[str, object],
+) -> bool:
+    """Whether the item must carry a blocking-reason / next-action comment (AC4).
+
+    True when blocked or when any hazard flag is present.
+    """
+    names = label_names(labels)
+    if "status:blocked" in names:
+        return True
+    return any(name.startswith("hazard:") for name in names)
+
+
+def labels_for_new_issue(
+    issue_type: str | None = None,
+    initial_status: str | None = None,
+    extra_labels: Sequence[str] | None = None,
+    *,
+    discussion: bool = False,
+) -> list[str]:
+    names = list(extra_labels or [])
+    if issue_type:
+        type_name = issue_type if issue_type.startswith("type:") else f"type:{issue_type}"
+        names.append(type_name)
+    if discussion:
+        names.append("type:discussion")
+    if initial_status:
+        names.append(canonical_status_label(initial_status))
+
+    result: list[str] = []
+    for name in names:
+        if name not in result:
+            result.append(name)
+    return result
+
+
+def assess_issue_labels(
+    labels: Iterable[str | Mapping[str, object]] | Mapping[str, object],
+    *,
+    discussion: bool = False,
+    require_type: bool = True,
+    require_status: bool = True,
+) -> dict:
+    names = label_names(labels)
+    found_types = type_labels(names)
+    found_statuses = status_labels(names)
+    found_roles = role_labels(names)
+    found_hazards = hazard_labels(names)
+    errors: list[str] = []
+    warnings: list[str] = []
+
+    if require_type and not found_types:
+        errors.append("issue is missing a type:* label")
+    if require_status and not found_statuses:
+        errors.append("issue is missing a status:* label")
+    if discussion and "type:discussion" not in found_types:
+        errors.append("discussion issue is missing type:discussion")
+    if len(found_statuses) > 1:
+        errors.append(
+            "issue has multiple active status:* labels: "
+            + ", ".join(found_statuses)
+        )
+    if len(found_roles) > 1:
+        errors.append(
+            "issue has multiple active role:* labels: " + ", ".join(found_roles)
+        )
+
+    for name in found_types:
+        if name not in TYPE_LABELS:
+            warnings.append(f"unknown type label '{name}'")
+    for name in found_statuses:
+        if name not in STATUS_LABELS:
+            warnings.append(f"unknown status label '{name}'")
+    for name in found_roles:
+        if name not in ROLE_LABELS:
+            warnings.append(f"unknown role label '{name}'")
+    for name in found_hazards:
+        if name not in HAZARD_LABELS:
+            warnings.append(f"unknown hazard label '{name}'")
+
+    return {
+        "valid": not errors,
+        "labels": names,
+        "type_labels": found_types,
+        "status_labels": found_statuses,
+        "role_labels": found_roles,
+        "hazard_labels": found_hazards,
+        "errors": errors,
+        "warnings": warnings,
+    }

lease_lifecycle.py

@@ -0,0 +1,723 @@
+"""First-class control-plane lease lifecycle (#601).
+
+Active leases are queryable workflow state. Canonical operations:
+
+* list / inspect
+* adopt (with provenance)
+* release (explicit, recorded)
+* expire / reclaim
+* abandon (requires proof: dead process and/or missing worktree, etc.)
+
+Authority model:
+
+* Control-plane DB is the coordination source for assignment/lease.
+* File locks and comment-only leases are **not** authoritative alone.
+* Reviewer/merger comment leases (``reviewer_pr_lease`` / merger adoption)
+  remain for Gitea-thread durability; they must not bypass DB assignment when
+  the control-plane path is in use.
+
+Fail closed on ambiguous ownership, active foreign leases, mismatched
+worktree, stale head, missing capability, and unsafe cleanup.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+from dataclasses import dataclass, field
+from datetime import datetime, timezone
+from typing import Any, Callable, Mapping
+
+import control_plane_db as cpd
+
+# Outcomes / safe next actions (stable vocabulary for tools + tests).
+SAFE_OWNER_RESUME = "owner_resume"
+SAFE_WAIT_FOREIGN = "wait_foreign_active"
+SAFE_RECLAIM_EXPIRED = "reclaim_expired"
+SAFE_ABANDON_ALLOWED = "abandon_allowed"
+SAFE_RELEASE_OWNED = "release_owned"
+SAFE_STALE_PROMPT = "stale_prompt_lease"
+SAFE_UNKNOWN = "inspect_only"
+SAFE_NO_AUTHORITY = "file_or_comment_not_authoritative"
+
+LEASE_STATUS_ACTIVE = "active"
+LEASE_STATUS_RELEASED = "released"
+LEASE_STATUS_EXPIRED = "expired"
+LEASE_STATUS_ABANDONED = "abandoned"
+
+AUTHORITATIVE_SOURCE = "control_plane_db"
+
+
+class LeaseLifecycleError(RuntimeError):
+    """Fail-closed lifecycle policy error."""
+
+
+@dataclass(frozen=True)
+class AbandonProof:
+    """Required evidence to abandon a non-owned or expired lease safely."""
+
+    dead_process: bool = False
+    missing_worktree: bool = False
+    same_owner: bool = False
+    no_open_pr: bool = False
+    no_live_mutation_risk: bool = False
+    operator_authorized: bool = False
+    worktree_path: str | None = None
+    owner_pid: int | None = None
+    notes: str = ""
+
+    def as_dict(self) -> dict[str, Any]:
+        return {
+            "dead_process": self.dead_process,
+            "missing_worktree": self.missing_worktree,
+            "same_owner": self.same_owner,
+            "no_open_pr": self.no_open_pr,
+            "no_live_mutation_risk": self.no_live_mutation_risk,
+            "operator_authorized": self.operator_authorized,
+            "worktree_path": self.worktree_path,
+            "owner_pid": self.owner_pid,
+            "notes": self.notes,
+        }
+
+    def is_sufficient(self) -> bool:
+        """Abandon requires process death or missing worktree + no mutation risk.
+
+        Foreign active leases additionally need operator_authorized unless the
+        owner process is dead and the worktree is missing.
+        """
+        if not self.no_live_mutation_risk:
+            return False
+        if not (self.dead_process or self.missing_worktree):
+            return False
+        if self.same_owner:
+            return True
+        # Foreign abandon: operator flag OR (dead + missing worktree + no risk)
+        if self.operator_authorized:
+            return True
+        return bool(self.dead_process and self.missing_worktree and self.no_open_pr)
+
+
+def is_process_alive(pid: int | None) -> bool:
+    if pid is None:
+        return False
+    try:
+        pid_i = int(pid)
+    except (TypeError, ValueError):
+        return False
+    if pid_i <= 0:
+        return False
+    try:
+        os.kill(pid_i, 0)
+        return True
+    except ProcessLookupError:
+        return False
+    except PermissionError:
+        # Exists but not owned by us — treat as alive (fail closed).
+        return True
+    except OSError:
+        return False
+
+
+def worktree_exists(path: str | None) -> bool:
+    if not path:
+        return False
+    try:
+        return os.path.isdir(os.path.realpath(path))
+    except OSError:
+        return False
+
+
+def _utc_now() -> datetime:
+    return datetime.now(timezone.utc)
+
+
+def _parse_ts(value: str | None) -> datetime | None:
+    return cpd._parse_ts(value)
+
+
+def classify_lease_freshness(
+    lease: Mapping[str, Any],
+    *,
+    now: datetime | None = None,
+    pid_checker: Callable[[int | None], bool] = is_process_alive,
+    worktree_checker: Callable[[str | None], bool] = worktree_exists,
+) -> dict[str, Any]:
+    """Classify a control-plane lease row for workflow tooling."""
+    moment = now or _utc_now()
+    status = (lease.get("status") or "").strip().lower()
+    expires = _parse_ts(lease.get("expires_at"))
+    owner_pid = lease.get("owner_pid")
+    if owner_pid is None:
+        owner_pid = lease.get("session_pid")
+    wt = lease.get("worktree_path")
+    pid_alive = pid_checker(owner_pid) if owner_pid is not None else None
+    wt_present = worktree_checker(wt) if wt else None
+    expired_by_time = bool(expires and expires <= moment)
+
+    if status == LEASE_STATUS_ABANDONED:
+        freshness = "abandoned"
+    elif status == LEASE_STATUS_RELEASED:
+        freshness = "released"
+    elif status == LEASE_STATUS_EXPIRED or expired_by_time:
+        freshness = "expired"
+    elif status != LEASE_STATUS_ACTIVE:
+        freshness = status or "unknown"
+    elif pid_alive is False:
+        freshness = "stale_dead_process"
+    elif wt is not None and wt_present is False:
+        freshness = "stale_missing_worktree"
+    else:
+        freshness = "active"
+
+    return {
+        "freshness": freshness,
+        "status": status,
+        "expired_by_time": expired_by_time,
+        "owner_pid": owner_pid,
+        "owner_pid_alive": pid_alive,
+        "worktree_path": wt,
+        "worktree_present": wt_present,
+        "expires_at": lease.get("expires_at"),
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+    }
+
+
+def decide_safe_next_action(
+    *,
+    lease: Mapping[str, Any] | None,
+    caller_session_id: str,
+    freshness: Mapping[str, Any] | None = None,
+    caller_worktree: str | None = None,
+) -> dict[str, Any]:
+    """Return safe_next_action + reasons for a caller inspecting a lease."""
+    if not lease:
+        return {
+            "safe_next_action": SAFE_STALE_PROMPT,
+            "reasons": ["lease not found in control-plane DB (stale prompt id?)"],
+            "block": True,
+        }
+    fr = freshness or classify_lease_freshness(lease)
+    owner = str(lease.get("session_id") or "")
+    same_owner = owner == str(caller_session_id)
+    status = fr.get("freshness")
+    reasons: list[str] = []
+
+    # Worktree mismatch for owner resume
+    lease_wt = lease.get("worktree_path")
+    if (
+        same_owner
+        and lease_wt
+        and caller_worktree
+        and os.path.realpath(str(lease_wt)) != os.path.realpath(str(caller_worktree))
+    ):
+        return {
+            "safe_next_action": SAFE_UNKNOWN,
+            "reasons": [
+                f"worktree mismatch: lease has {lease_wt!r}, caller has "
+                f"{caller_worktree!r} (fail closed)"
+            ],
+            "block": True,
+            "same_owner": True,
+        }
+
+    if status in ("abandoned", "released"):
+        return {
+            "safe_next_action": SAFE_STALE_PROMPT,
+            "reasons": [f"lease status is {status}; do not adopt blindly"],
+            "block": True,
+            "same_owner": same_owner,
+        }
+
+    if status == "expired" or fr.get("expired_by_time"):
+        return {
+            "safe_next_action": SAFE_RECLAIM_EXPIRED,
+            "reasons": ["lease expired; reclaim via expire+assign or adopt reclaim path"],
+            "block": False,
+            "same_owner": same_owner,
+        }
+
+    if status in ("stale_dead_process", "stale_missing_worktree"):
+        if same_owner:
+            return {
+                "safe_next_action": SAFE_OWNER_RESUME,
+                "reasons": [
+                    f"caller owns lease with freshness={status}; rebind via "
+                    "adopt/owner-resume or abandon with proof"
+                ],
+                "block": False,
+                "same_owner": True,
+                "also_allowed": [SAFE_ABANDON_ALLOWED, SAFE_RELEASE_OWNED],
+            }
+        return {
+            "safe_next_action": SAFE_ABANDON_ALLOWED,
+            "reasons": [
+                f"lease freshness={status}; abandon with required proof then reassign"
+            ],
+            "block": False,
+            "same_owner": False,
+        }
+
+    if same_owner and status == "active":
+        return {
+            "safe_next_action": SAFE_OWNER_RESUME,
+            "reasons": [
+                "caller owns active lease; resume via heartbeat/assign owner-resume "
+                "or release explicitly"
+            ],
+            "block": False,
+            "same_owner": True,
+            "also_allowed": [SAFE_RELEASE_OWNED],
+        }
+
+    if not same_owner and status == "active":
+        return {
+            "safe_next_action": SAFE_WAIT_FOREIGN,
+            "reasons": [
+                f"foreign active lease held by session {owner}; "
+                "do not steal (fail closed)"
+            ],
+            "block": True,
+            "same_owner": False,
+            "owner_session_id": owner,
+        }
+
+    return {
+        "safe_next_action": SAFE_UNKNOWN,
+        "reasons": [f"unclassified freshness={status}"],
+        "block": True,
+        "same_owner": same_owner,
+    }
+
+
+def non_db_lease_authority_report(
+    *,
+    file_lock_present: bool = False,
+    comment_lease_present: bool = False,
+    db_lease_present: bool = False,
+) -> dict[str, Any]:
+    """File/comment leases alone are not coordination authority (#601 / #600)."""
+    authoritative = bool(db_lease_present)
+    reasons: list[str] = []
+    if file_lock_present and not db_lease_present:
+        reasons.append(
+            "file lock present but control-plane DB lease absent — "
+            "file lock is not authoritative alone"
+        )
+    if comment_lease_present and not db_lease_present:
+        reasons.append(
+            "comment-only lease present but control-plane DB lease absent — "
+            "comment lease is not authoritative alone"
+        )
+    if authoritative:
+        reasons.append("control-plane DB lease is the coordination authority")
+    return {
+        "authoritative_source": AUTHORITATIVE_SOURCE if authoritative else None,
+        "db_lease_present": db_lease_present,
+        "file_lock_present": file_lock_present,
+        "comment_lease_present": comment_lease_present,
+        "safe_next_action": (
+            SAFE_UNKNOWN if authoritative else SAFE_NO_AUTHORITY
+        ),
+        "reasons": reasons,
+        "file_lock_only": bool(file_lock_present and not db_lease_present),
+        "comment_lease_only": bool(comment_lease_present and not db_lease_present),
+    }
+
+
+def build_adopt_provenance(
+    *,
+    adopted_from_session_id: str,
+    adopted_by_session_id: str,
+    work_kind: str,
+    work_number: int,
+    remote: str,
+    org: str,
+    repo: str,
+    worktree_path: str | None,
+    expected_head_sha: str | None,
+    prior_lease_id: str | None,
+    reason: str,
+) -> dict[str, Any]:
+    return {
+        "adopted_from_session_id": adopted_from_session_id,
+        "adopted_by_session_id": adopted_by_session_id,
+        "work_kind": work_kind,
+        "work_number": int(work_number),
+        "remote": remote,
+        "org": org,
+        "repo": repo,
+        "worktree_path": worktree_path,
+        "expected_head_sha": expected_head_sha,
+        "prior_lease_id": prior_lease_id,
+        "reason": reason,
+        "recorded_at": cpd._ts(),
+        "source": "lease_lifecycle.adopt",
+    }
+
+
+def inspect_lease(
+    db: cpd.ControlPlaneDB,
+    lease_id: str,
+    *,
+    caller_session_id: str,
+    caller_worktree: str | None = None,
+) -> dict[str, Any]:
+    """Full first-class lease workflow state for one lease id."""
+    state = db.get_lease_workflow_state(lease_id)
+    if not state:
+        decision = decide_safe_next_action(
+            lease=None, caller_session_id=caller_session_id
+        )
+        return {
+            "success": True,
+            "found": False,
+            "lease_id": lease_id,
+            "lease": None,
+            "freshness": None,
+            **decision,
+            "authoritative_source": AUTHORITATIVE_SOURCE,
+            "file_lock_only": False,
+            "comment_lease_only": False,
+        }
+    lease = state["lease"]
+    freshness = classify_lease_freshness(lease)
+    decision = decide_safe_next_action(
+        lease=lease,
+        caller_session_id=caller_session_id,
+        freshness=freshness,
+        caller_worktree=caller_worktree,
+    )
+    return {
+        "success": True,
+        "found": True,
+        "lease_id": lease_id,
+        "lease": lease,
+        "assignment": state.get("assignment"),
+        "work_item": state.get("work_item"),
+        "session": state.get("session"),
+        "freshness": freshness,
+        "provenance": state.get("provenance"),
+        **decision,
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+    }
+
+
+def list_active_leases(
+    db: cpd.ControlPlaneDB,
+    *,
+    remote: str | None = None,
+    org: str | None = None,
+    repo: str | None = None,
+    role: str | None = None,
+    include_non_active: bool = False,
+    limit: int = 100,
+) -> dict[str, Any]:
+    statuses = None if include_non_active else (LEASE_STATUS_ACTIVE,)
+    rows = db.list_leases(
+        remote=remote,
+        org=org,
+        repo=repo,
+        role=role,
+        statuses=statuses,
+        limit=limit,
+    )
+    enriched = []
+    for row in rows:
+        fr = classify_lease_freshness(row)
+        enriched.append({**row, "freshness": fr})
+    return {
+        "success": True,
+        "count": len(enriched),
+        "leases": enriched,
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+        "include_non_active": include_non_active,
+    }
+
+
+def adopt_lease(
+    db: cpd.ControlPlaneDB,
+    *,
+    lease_id: str,
+    adopter_session_id: str,
+    role: str,
+    worktree_path: str | None = None,
+    expected_head_sha: str | None = None,
+    owner_pid: int | None = None,
+    operator_authorized: bool = False,
+) -> dict[str, Any]:
+    """Sanctioned adopt path with provenance; never silent foreign steal."""
+    state = db.get_lease_workflow_state(lease_id)
+    if not state:
+        raise LeaseLifecycleError(
+            f"unknown lease_id {lease_id}; stale prompt id (fail closed)"
+        )
+    lease = state["lease"]
+    work = state["work_item"]
+    freshness = classify_lease_freshness(lease)
+    owner = str(lease.get("session_id") or "")
+    same_owner = owner == str(adopter_session_id)
+
+    if freshness["freshness"] == "active" and not same_owner:
+        raise LeaseLifecycleError(
+            f"refusing to steal active foreign lease {lease_id} owned by "
+            f"{owner} (fail closed)"
+        )
+
+    if freshness["freshness"] in ("abandoned", "released"):
+        raise LeaseLifecycleError(
+            f"lease {lease_id} is {freshness['freshness']}; cannot adopt "
+            "(fail closed)"
+        )
+
+    # Expired or stale: require abandon-style safety before ownership transfer
+    # when not same owner; same owner may reclaim.
+    if not same_owner and freshness["freshness"] in (
+        "expired",
+        "stale_dead_process",
+        "stale_missing_worktree",
+    ):
+        if not operator_authorized and freshness["freshness"] == "expired":
+            # Deterministic reclaim of expired foreign lease is allowed
+            # without operator flag (sanctioned expire reclaim).
+            pass
+        elif freshness["freshness"] != "expired" and not operator_authorized:
+            # stale active-looking requires abandon proof path
+            raise LeaseLifecycleError(
+                f"lease {lease_id} freshness={freshness['freshness']}; "
+                "use abandon with proof before foreign adopt (fail closed)"
+            )
+
+    provenance = build_adopt_provenance(
+        adopted_from_session_id=owner,
+        adopted_by_session_id=adopter_session_id,
+        work_kind=str(work.get("kind")),
+        work_number=int(work.get("number")),
+        remote=str(work.get("remote")),
+        org=str(work.get("org")),
+        repo=str(work.get("repo")),
+        worktree_path=worktree_path,
+        expected_head_sha=expected_head_sha or lease.get("expected_head_sha"),
+        prior_lease_id=lease_id,
+        reason=(
+            "owner-resume-adopt" if same_owner else "sanctioned-reclaim-adopt"
+        ),
+    )
+
+    result = db.adopt_lease(
+        lease_id=lease_id,
+        adopter_session_id=adopter_session_id,
+        role=role,
+        worktree_path=worktree_path,
+        expected_head_sha=expected_head_sha,
+        owner_pid=owner_pid if owner_pid is not None else os.getpid(),
+        provenance=provenance,
+    )
+    return {
+        "success": True,
+        "outcome": result.get("outcome"),
+        "same_owner": same_owner,
+        "prior_lease_id": lease_id,
+        "lease": result.get("lease"),
+        "assignment": result.get("assignment"),
+        "provenance": provenance,
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+        "reasons": result.get("reasons") or [],
+    }
+
+
+def release_lease(
+    db: cpd.ControlPlaneDB,
+    *,
+    lease_id: str,
+    session_id: str,
+) -> dict[str, Any]:
+    proof = db.release_lease_recorded(lease_id=lease_id, session_id=session_id)
+    return {
+        "success": True,
+        "outcome": "released",
+        "lease_id": lease_id,
+        "session_id": session_id,
+        "release_proof": proof,
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+    }
+
+
+def expire_leases(db: cpd.ControlPlaneDB) -> dict[str, Any]:
+    n = db.expire_stale_leases()
+    return {
+        "success": True,
+        "outcome": "expired",
+        "expired_count": int(n),
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+    }
+
+
+def reclaim_expired_lease(
+    db: cpd.ControlPlaneDB,
+    *,
+    lease_id: str,
+    session_id: str,
+    role: str,
+    worktree_path: str | None = None,
+    expected_head_sha: str | None = None,
+) -> dict[str, Any]:
+    """Expire-if-needed then assign to *session_id* for the same work item."""
+    state = db.get_lease_workflow_state(lease_id)
+    if not state:
+        raise LeaseLifecycleError(f"unknown lease_id {lease_id}")
+    lease = state["lease"]
+    work = state["work_item"]
+    freshness = classify_lease_freshness(lease)
+    if freshness["freshness"] == "active":
+        if lease.get("session_id") != session_id:
+            raise LeaseLifecycleError(
+                f"cannot reclaim active foreign lease {lease_id} (fail closed)"
+            )
+        # owner resume
+        return adopt_lease(
+            db,
+            lease_id=lease_id,
+            adopter_session_id=session_id,
+            role=role,
+            worktree_path=worktree_path,
+            expected_head_sha=expected_head_sha,
+        )
+    if freshness["freshness"] not in (
+        "expired",
+        "stale_dead_process",
+        "stale_missing_worktree",
+        "abandoned",
+        "released",
+    ):
+        raise LeaseLifecycleError(
+            f"lease {lease_id} freshness={freshness['freshness']} not reclaimable"
+        )
+
+    # Ensure expired marker is applied
+    db.expire_stale_leases()
+    # If still active due to clock skew / non-time stale, force expire this lease
+    refreshed = db.get_lease_workflow_state(lease_id)
+    if refreshed and refreshed["lease"].get("status") == "active":
+        db.force_expire_lease(lease_id, reason="reclaim_expired_lease")
+
+    head = expected_head_sha or work.get("current_head_sha")
+    assigned = db.assign_and_lease(
+        session_id=session_id,
+        role=role,
+        remote=str(work["remote"]),
+        org=str(work["org"]),
+        repo=str(work["repo"]),
+        kind=str(work["kind"]),
+        number=int(work["number"]),
+        expected_head_sha=head,
+        phase="reclaimed",
+        worktree_path=worktree_path,
+        owner_pid=os.getpid(),
+    )
+    if assigned.outcome != "assigned":
+        raise LeaseLifecycleError(
+            f"reclaim assign failed: {assigned.outcome} — {assigned.reason}"
+        )
+    provenance = build_adopt_provenance(
+        adopted_from_session_id=str(lease.get("session_id")),
+        adopted_by_session_id=session_id,
+        work_kind=str(work["kind"]),
+        work_number=int(work["number"]),
+        remote=str(work["remote"]),
+        org=str(work["org"]),
+        repo=str(work["repo"]),
+        worktree_path=worktree_path,
+        expected_head_sha=head,
+        prior_lease_id=lease_id,
+        reason="reclaim_expired",
+    )
+    db.attach_lease_provenance(assigned.lease_id, provenance)
+    return {
+        "success": True,
+        "outcome": "reclaimed",
+        "prior_lease_id": lease_id,
+        "assignment": assigned.as_dict(),
+        "provenance": provenance,
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+    }
+
+
+def abandon_lease(
+    db: cpd.ControlPlaneDB,
+    *,
+    lease_id: str,
+    requester_session_id: str,
+    proof: AbandonProof,
+) -> dict[str, Any]:
+    state = db.get_lease_workflow_state(lease_id)
+    if not state:
+        raise LeaseLifecycleError(f"unknown lease_id {lease_id}")
+    lease = state["lease"]
+    owner = str(lease.get("session_id") or "")
+    same_owner = owner == str(requester_session_id)
+
+    # Enrich proof with live checks when not provided
+    owner_pid = proof.owner_pid
+    if owner_pid is None:
+        owner_pid = lease.get("owner_pid")
+    wt = proof.worktree_path if proof.worktree_path is not None else lease.get(
+        "worktree_path"
+    )
+    dead = proof.dead_process or (
+        owner_pid is not None and not is_process_alive(owner_pid)
+    )
+    missing_wt = proof.missing_worktree or (
+        bool(wt) and not worktree_exists(str(wt))
+    )
+    enriched = AbandonProof(
+        dead_process=bool(dead),
+        missing_worktree=bool(missing_wt),
+        same_owner=same_owner or proof.same_owner,
+        no_open_pr=proof.no_open_pr,
+        no_live_mutation_risk=proof.no_live_mutation_risk,
+        operator_authorized=proof.operator_authorized,
+        worktree_path=str(wt) if wt else None,
+        owner_pid=int(owner_pid) if owner_pid is not None else None,
+        notes=proof.notes,
+    )
+    if not enriched.is_sufficient():
+        raise LeaseLifecycleError(
+            "abandon proof insufficient: need (dead_process or missing_worktree) "
+            "and no_live_mutation_risk; foreign abandon also needs "
+            "operator_authorized or (dead+missing_worktree+no_open_pr). "
+            f"proof={enriched.as_dict()}"
+        )
+
+    # Active foreign with live process + present worktree: never abandon without
+    # operator (already covered by is_sufficient).
+    result = db.abandon_lease(
+        lease_id=lease_id,
+        requester_session_id=requester_session_id,
+        proof=enriched.as_dict(),
+    )
+    return {
+        "success": True,
+        "outcome": "abandoned",
+        "lease_id": lease_id,
+        "requester_session_id": requester_session_id,
+        "prior_owner_session_id": owner,
+        "abandon_proof": enriched.as_dict(),
+        "audit": result,
+        "authoritative_source": AUTHORITATIVE_SOURCE,
+        "file_lock_only": False,
+        "comment_lease_only": False,
+    }

manage_labels.py

@@ -22,7 +22,8 @@ venv_python = os.path.join(PROJECT_ROOT, "venv", "bin", "python3")
 if os.path.exists(venv_python) and sys.executable != venv_python:
     os.execv(venv_python, [venv_python] + sys.argv)
 
-from gitea_auth import get_auth_header, api_request, repo_api_url
+from gitea_auth import get_auth_header, api_request, api_get_all, repo_api_url
+import issue_workflow_labels
 
 HOST = "gitea.dadeschools.net"
 ORG = "Contractor"
@@ -35,8 +36,10 @@ LABELS = [
     {"name": "epic", "color": "8250df", "description": ""},
     {"name": "important", "color": "fbca04", "description": ""},
     {"name": "nice-to-have", "color": "0e8a16", "description": ""},
-    {"name": "status:in-progress", "color": "fefe2e",
-     "description": "Issue is being worked on"},
+    *[
+        {"name": spec.name, "color": spec.color, "description": spec.description}
+        for spec in issue_workflow_labels.CANONICAL_LABEL_SPECS
+    ],
 ]
 
 # issue number -> label names to apply (one-off backfill)
@@ -79,9 +82,17 @@ def api(method, path, auth, payload=None):
 
 
 def _labels_by_name(auth):
-    """Return {label name: id} for the repo's existing labels."""
-    existing = api("GET", "/labels?limit=100", auth) or []
-    return {lb["name"]: lb["id"] for lb in existing}
+    """Return {label name: id} for the repo's existing labels (all pages, #627)."""
+    existing = api_get_all(f"{BASE_URL}/labels", auth) or []
+    name_to_id = {}
+    for lb in existing:
+        if not isinstance(lb, dict):
+            continue
+        name = lb.get("name")
+        lid = lb.get("id")
+        if name and lid is not None and name not in name_to_id:
+            name_to_id[name] = lid
+    return name_to_id
 
 
 def create_labels(auth, dry=False):

mark_issue.py

@@ -17,7 +17,7 @@ if os.path.exists(venv_python) and sys.executable != venv_python:
 
 from gitea_auth import (
     get_auth_header, resolve_remote, add_remote_args,
-    api_request, repo_api_url,
+    api_request, api_get_all, repo_api_url,
 )
 
 LABEL_NAME = "status:in-progress"
@@ -45,12 +45,12 @@ def main(argv=None):
     base = repo_api_url(host, org, repo)
 
     try:
-        # Find the label ID
-        labels = api_request("GET", f"{base}/labels?limit=100", auth)
+        # Paginated inventory (#627): Gitea caps single pages at 50.
+        labels = api_get_all(f"{base}/labels", auth) or []
         label_id = None
         for lb in labels:
-            if lb["name"] == LABEL_NAME:
-                label_id = lb["id"]
+            if lb.get("name") == LABEL_NAME:
+                label_id = lb.get("id")
                 break
 
         if label_id is None:

mcp-menu.sh

@@ -115,7 +115,15 @@ Workflow:
 }
 
 show_reviewer_prompts() {
-  print_prompt_block "Reviewer — PR review" \
+  while true; do
+    printf '\n--- Reviewer workflow prompts ---\n'
+    printf '  1) Standard PR review\n'
+    printf '  2) Skip already-reviewed stale REQUEST_CHANGES PR and hand off to author\n'
+    printf '  0) Back\n'
+    read -r -p 'Choice: ' choice
+    case "$choice" in
+      1)
+        print_prompt_block "Reviewer — PR review" \
 "You are the REVIEWER session for <org>/<repo>.
 
 Goal: review PR #<P> only — do not merge unless explicitly switched to merger mode.
@@ -126,6 +134,35 @@ Workflow:
 3. gitea_view_pr, validate scope, run required checks in the correct worktree.
 4. gitea_review_pr with approve, request-changes, or comment as warranted.
 5. Final report: PR head SHA, verdict, validation evidence, mutation ledger. No merge in reviewer-only runs."
+        ;;
+      2)
+        print_prompt_block "Reviewer — skip already-reviewed stale REQUEST_CHANGES PR and hand off to author" \
+"You are the REVIEWER session for <org>/<repo>.
+
+Goal: review PR #<P> only far enough to determine whether the current head already has a non-stale REQUEST_CHANGES verdict. If it does, do NOT submit another terminal review mutation — produce an author handoff and stop.
+
+Workflow:
+1. gitea_view_pr; pin the current head SHA.
+2. Load prior reviews; find the latest REQUEST_CHANGES and the head SHA it was bound to.
+3. If that REQUEST_CHANGES is still bound to the current head (non-stale), do not submit another terminal review mutation. Confirm the binding and summarize the blockers.
+4. Run only the diagnostics needed to produce a useful author handoff — no full re-review.
+5. Duplicate/supersession check: confirm no newer canonical PR or superseding head changes the decision.
+6. Stop — no new review mutation.
+
+Return:
+- selected PR and issue
+- current head SHA
+- prior REQUEST_CHANGES proof (review id + bound head SHA)
+- duplicate/supersession analysis
+- validation summary
+- why no new review mutation was submitted
+- corrected mutation ledger, including local worktree create/remove if used
+- author-ready fix prompt"
+        ;;
+      0) return ;;
+      *) printf 'Invalid choice.\n' ;;
+    esac
+  done
 }
 
 show_merger_prompts() {

mcp_daemon_guard.py

@@ -0,0 +1,517 @@
+"""Sanctioned MCP daemon guards for imports and credential access (#558 / #695).
+
+Direct ``import gitea_mcp_server`` from a shell bypasses native MCP transport.
+#558 introduced a daemon marker; #695 hardens it so:
+
+- Environment variables alone cannot reconstruct a native session
+  (``GITEA_MCP_SANCTIONED_DAEMON=1`` / ``GITEA_ALLOW_DIRECT_MCP_IMPORT=1`` are
+  insufficient for mutation gates).
+- A process-local runtime record is established only by the resolved canonical
+  entrypoint path (not basename) **and** the actual native MCP transport
+  lifecycle (``bind_native_mcp_transport`` before ``mcp.run``). Merely
+  importing or launching the entrypoint offline does not grant mutation
+  authority.
+- Public caller-controlled flags (including any former
+  ``allow_test_bootstrap``) never establish trusted mutation provenance.
+- Offline scripts that import internals fail closed on mutations.
+- Pytest remains allowed for hermetic unit tests via ``is_pytest_runtime()``.
+  A separate test-only seam may establish a **test-mode** native record for
+  unit tests of transport gates; that record cannot authorize production
+  Gitea mutation endpoints.
+
+Manual deletion of session-state files is never a recovery path.
+"""
+
+from __future__ import annotations
+
+import hashlib
+import inspect
+import os
+import secrets
+import time
+from pathlib import Path
+from typing import Any
+
+SANCTIONED_DAEMON_ENV = "GITEA_MCP_SANCTIONED_DAEMON"
+ALLOW_DIRECT_IMPORT_ENV = "GITEA_ALLOW_DIRECT_MCP_IMPORT"
+ALLOW_KEYCHAIN_CLI_ENV = "GITEA_ALLOW_KEYCHAIN_CLI"
+# Test-only: force provenance failure even under pytest (#695 regressions).
+FORCE_PROVENANCE_FAIL_ENV = "GITEA_TEST_FORCE_UNSANCTIONED"
+
+# Process-local native runtime (never persisted, never read from env alone).
+_NATIVE_RUNTIME: dict[str, Any] | None = None
+
+# Production transport identifiers accepted by bind_native_mcp_transport.
+_PRODUCTION_TRANSPORTS = frozenset({"stdio"})
+_RUNTIME_MODE_PRODUCTION = "production"
+_RUNTIME_MODE_TEST = "test"
+_PHASE_ENTRYPOINT_CLAIMED = "entrypoint_claimed"
+_PHASE_TRANSPORT_BOUND = "transport_bound"
+
+# Default session-state root (mirrors mcp_session_state; kept local to avoid
+# import cycles). Used only to pin authority at transport bind (#695 AC2).
+_DEFAULT_SESSION_STATE_DIR = os.path.expanduser("~/.cache/gitea-tools/session-state")
+SESSION_STATE_DIR_ENV = "GITEA_MCP_SESSION_STATE_DIR"
+
+
+class UnsanctionedRuntimeError(RuntimeError):
+    """Raised when mutation/credential code runs outside a native MCP daemon."""
+
+
+def is_pytest_runtime() -> bool:
+    if (os.environ.get(FORCE_PROVENANCE_FAIL_ENV) or "").strip() in {
+        "1",
+        "true",
+        "yes",
+    }:
+        return False
+    import sys
+
+    if "pytest" in sys.modules:
+        return True
+    return bool((os.environ.get("PYTEST_CURRENT_TEST") or "").strip())
+
+
+def _package_root() -> Path:
+    """Directory that contains the canonical MCP entrypoint modules."""
+    return Path(__file__).resolve().parent
+
+
+def canonical_entrypoint_paths() -> frozenset[str]:
+    """Resolved absolute paths of official entrypoints (not basenames)."""
+    root = _package_root()
+    return frozenset(
+        {
+            str((root / "mcp_server.py").resolve()),
+            str((root / "gitea_mcp_server.py").resolve()),
+        }
+    )
+
+
+def _resolve_path(path: str | None) -> str | None:
+    if not path:
+        return None
+    try:
+        return str(Path(path).resolve())
+    except (OSError, RuntimeError, ValueError):
+        return None
+
+
+def _caller_official_entrypoint_path() -> str | None:
+    """Return the resolved canonical entrypoint path in the call stack, or None.
+
+    Basename-only matches (e.g. an attacker file named ``mcp_server.py``
+    elsewhere) are rejected. The path must equal one of
+    :func:`canonical_entrypoint_paths`.
+    """
+    canonical = canonical_entrypoint_paths()
+    for frame in inspect.stack()[1:20]:
+        resolved = _resolve_path(frame.filename)
+        if resolved and resolved in canonical:
+            return resolved
+    return None
+
+
+def _caller_is_official_entrypoint() -> bool:
+    """True when invoked from a resolved canonical entrypoint path (#695)."""
+    return _caller_official_entrypoint_path() is not None
+
+
+def _new_runtime_token() -> tuple[str, str]:
+    token = secrets.token_hex(32)
+    fingerprint = hashlib.sha256(token.encode()).hexdigest()[:16]
+    return token, fingerprint
+
+
+def mark_sanctioned_daemon() -> dict[str, Any]:
+    """Claim the official entrypoint for this process (#695).
+
+    This alone does **not** authorize mutations. Callers must subsequently
+    bind the native MCP transport via :func:`bind_native_mcp_transport`.
+
+    Only a stack frame whose **resolved absolute path** is the canonical
+    ``mcp_server.py`` or ``gitea_mcp_server.py`` next to this module may
+    claim the entrypoint. Basename spoofing is rejected.
+
+    There is no public ``allow_test_bootstrap`` argument: caller-controlled
+    flags must never establish trusted mutation provenance. Hermetic tests
+    use :func:`install_test_native_runtime` (pytest-only, test mode).
+    """
+    global _NATIVE_RUNTIME
+    if is_pytest_runtime():
+        # Under pytest, production mark is a no-op for transport authority.
+        # Tests that need a native-transport record use install_test_native_runtime.
+        return native_runtime_status()
+
+    entrypoint_path = _caller_official_entrypoint_path()
+    if entrypoint_path is None:
+        raise UnsanctionedRuntimeError(
+            "mark_sanctioned_daemon rejected: not called from the resolved "
+            "canonical MCP entrypoint path (#695). Basename-only names "
+            "(e.g. a renamed runner called mcp_server.py) are insufficient. "
+            "Offline import / standalone scripts cannot reconstruct native "
+            "transport. Stop after native MCP failure; do not run offline "
+            "mutation helpers."
+        )
+
+    token, fingerprint = _new_runtime_token()
+    _NATIVE_RUNTIME = {
+        "token": token,
+        "token_fingerprint": fingerprint,
+        "pid": os.getpid(),
+        "started_at": time.time(),
+        "entrypoint": "mcp_server",
+        "entrypoint_path": entrypoint_path,
+        "phase": _PHASE_ENTRYPOINT_CLAIMED,
+        "transport": None,
+        "mode": _RUNTIME_MODE_PRODUCTION,
+    }
+    # Legacy signal for older probes; alone does not authorize mutations.
+    os.environ[SANCTIONED_DAEMON_ENV] = "1"
+    return native_runtime_status()
+
+
+def bind_native_mcp_transport(*, transport: str) -> dict[str, Any]:
+    """Bind the live native MCP transport lifecycle (#695).
+
+    Must be called from the resolved canonical entrypoint immediately before
+    the real MCP server transport loop (e.g. ``mcp.run(transport=\"stdio\")``).
+    Requires a prior successful :func:`mark_sanctioned_daemon` claim in this
+    process. Import-only or offline launch without this bind leaves
+    :func:`is_native_mcp_transport` false.
+    """
+    global _NATIVE_RUNTIME
+    transport_name = (transport or "").strip().lower()
+    if transport_name not in _PRODUCTION_TRANSPORTS:
+        raise UnsanctionedRuntimeError(
+            f"bind_native_mcp_transport rejected: transport {transport!r} is "
+            f"not a production MCP transport (#695). Allowed: "
+            f"{sorted(_PRODUCTION_TRANSPORTS)}."
+        )
+
+    entrypoint_path = _caller_official_entrypoint_path()
+    if entrypoint_path is None:
+        raise UnsanctionedRuntimeError(
+            "bind_native_mcp_transport rejected: not called from the resolved "
+            "canonical MCP entrypoint path (#695)."
+        )
+
+    if _NATIVE_RUNTIME is None or int(_NATIVE_RUNTIME.get("pid") or -1) != os.getpid():
+        raise UnsanctionedRuntimeError(
+            "bind_native_mcp_transport rejected: no entrypoint claim in this "
+            "process (#695). Call mark_sanctioned_daemon() from the official "
+            "entrypoint first."
+        )
+
+    if _NATIVE_RUNTIME.get("mode") != _RUNTIME_MODE_PRODUCTION:
+        raise UnsanctionedRuntimeError(
+            "bind_native_mcp_transport rejected: runtime mode is not "
+            "production (#695)."
+        )
+
+    claimed = (_NATIVE_RUNTIME.get("entrypoint_path") or "").strip()
+    if claimed and claimed != entrypoint_path:
+        raise UnsanctionedRuntimeError(
+            "bind_native_mcp_transport rejected: entrypoint path mismatch "
+            "between mark and bind (#695)."
+        )
+
+    # Pin session-state root for this server lifetime (#695 AC2 / PR #701).
+    # Changing GITEA_MCP_SESSION_STATE_DIR after bind must not manufacture a
+    # second authority domain for decision locks / workflow proofs.
+    raw_state = (os.environ.get(SESSION_STATE_DIR_ENV) or "").strip()
+    if not raw_state:
+        raw_state = _DEFAULT_SESSION_STATE_DIR
+    try:
+        pinned_state = str(Path(raw_state).resolve())
+    except (OSError, RuntimeError, ValueError):
+        pinned_state = raw_state
+
+    _NATIVE_RUNTIME["phase"] = _PHASE_TRANSPORT_BOUND
+    _NATIVE_RUNTIME["transport"] = transport_name
+    _NATIVE_RUNTIME["entrypoint_path"] = entrypoint_path
+    _NATIVE_RUNTIME["bound_at"] = time.time()
+    _NATIVE_RUNTIME["session_state_dir"] = pinned_state
+    os.environ[SANCTIONED_DAEMON_ENV] = "1"
+    return native_runtime_status()
+
+
+def install_test_native_runtime() -> dict[str, Any]:
+    """Pytest-only seam for hermetic native-transport unit tests (#695).
+
+    Establishes a **test-mode** process-local record so unit tests can exercise
+    gates that require ``is_native_mcp_transport()``. This record:
+
+    - is rejected outside pytest (including a fresh offline interpreter);
+    - never uses production mode;
+    - cannot authorize production Gitea mutation endpoints
+      (:func:`assert_production_mutation_runtime` / production path of
+      :func:`assert_sanctioned_mutation_runtime` when not under pytest).
+
+    There is no public caller-controlled flag that forges production native
+    transport.
+    """
+    global _NATIVE_RUNTIME
+    if not is_pytest_runtime():
+        raise UnsanctionedRuntimeError(
+            "install_test_native_runtime rejected: test-mode native runtime "
+            "is only available under pytest (#695). allow_test_bootstrap and "
+            "similar caller-controlled flags do not exist and cannot authorize "
+            "a fresh offline interpreter."
+        )
+    token, fingerprint = _new_runtime_token()
+    _NATIVE_RUNTIME = {
+        "token": token,
+        "token_fingerprint": fingerprint,
+        "pid": os.getpid(),
+        "started_at": time.time(),
+        "entrypoint": "test_bootstrap",
+        "entrypoint_path": None,
+        "phase": _PHASE_TRANSPORT_BOUND,
+        "transport": "test",
+        "mode": _RUNTIME_MODE_TEST,
+        "bound_at": time.time(),
+    }
+    return native_runtime_status()
+
+
+def clear_native_runtime_for_tests() -> None:
+    """Test helper: drop native runtime (does not clear env)."""
+    global _NATIVE_RUNTIME
+    _NATIVE_RUNTIME = None
+
+
+def pinned_session_state_dir() -> str | None:
+    """Session-state root pinned for this production transport lifetime (#695 AC2).
+
+    When production native transport is bound, durable session proofs must use
+    this directory only. Env overrides of ``GITEA_MCP_SESSION_STATE_DIR`` after
+    bind are ignored so redirected dirs (e.g. ``.mcp_session_701``) cannot
+    manufacture independent decision-lock authority (PR #701 recurrence).
+    """
+    if not is_production_native_mcp_transport():
+        return None
+    pinned = (_NATIVE_RUNTIME or {}).get("session_state_dir")
+    text = (str(pinned) if pinned is not None else "").strip()
+    return text or None
+
+
+def direct_import_env_enabled() -> bool:
+    """True when the legacy direct-import opt-in env is set (never authorizes)."""
+    return (os.environ.get(ALLOW_DIRECT_IMPORT_ENV) or "").strip().lower() in {
+        "1",
+        "true",
+        "yes",
+    }
+
+
+def assert_no_direct_import_bypass(context: str = "mutation") -> None:
+    """Fail closed when GITEA_ALLOW_DIRECT_MCP_IMPORT is used for mutations (#695 AC1).
+
+    The env flag is never a sanctioned recovery path for LLM/agent sessions.
+    Under pytest hermetic tests this is a no-op so unit tests can set the flag
+    to prove it does not grant authority.
+    """
+    if is_pytest_runtime():
+        return
+    if not direct_import_env_enabled():
+        return
+    raise UnsanctionedRuntimeError(
+        f"{ALLOW_DIRECT_IMPORT_ENV} does not authorize {context} (#695 AC1). "
+        "Direct import of gitea_mcp_server mutation tools is forbidden. "
+        "Stop after native MCP failure; reconnect the official MCP daemon. "
+        "Do not set direct-import flags, offline runners, or redirected "
+        f"{SESSION_STATE_DIR_ENV} directories to reconstruct gates."
+    )
+
+
+def is_native_mcp_transport() -> bool:
+    """True when this process holds a transport-bound native runtime (#695)."""
+    if (os.environ.get(FORCE_PROVENANCE_FAIL_ENV) or "").strip() in {
+        "1",
+        "true",
+        "yes",
+    }:
+        return False
+    if _NATIVE_RUNTIME is None:
+        return False
+    if int(_NATIVE_RUNTIME.get("pid") or -1) != os.getpid():
+        return False
+    if not (_NATIVE_RUNTIME.get("token") or "").strip():
+        return False
+    if _NATIVE_RUNTIME.get("phase") != _PHASE_TRANSPORT_BOUND:
+        return False
+    if not (_NATIVE_RUNTIME.get("transport") or "").strip():
+        return False
+    return True
+
+
+def is_production_native_mcp_transport() -> bool:
+    """True only for production-mode, transport-bound native runtime."""
+    if not is_native_mcp_transport():
+        return False
+    return (_NATIVE_RUNTIME or {}).get("mode") == _RUNTIME_MODE_PRODUCTION
+
+
+def is_sanctioned_mcp_daemon() -> bool:
+    """Backward-compatible name; #695 requires native transport, not env alone."""
+    if is_production_native_mcp_transport():
+        return True
+    if is_pytest_runtime():
+        return True
+    # Explicit direct-import override is for non-LLM operator/test tools only.
+    # It is deliberately ignored when a native runtime is expected for mutations
+    # under LLM sessions (tests use pytest path). Env alone never grants native.
+    return False
+
+
+def assert_production_mutation_runtime(context: str = "mutation") -> None:
+    """Fail closed unless production native MCP transport is bound (#695).
+
+    Test-mode bootstrap records and pytest-only hermetic allowances do **not**
+    satisfy this gate. Use for production Gitea mutation endpoints that must
+    never be reachable via test bootstrap.
+    """
+    if is_production_native_mcp_transport():
+        return
+    mode = (_NATIVE_RUNTIME or {}).get("mode")
+    if mode == _RUNTIME_MODE_TEST:
+        raise UnsanctionedRuntimeError(
+            f"Test-mode native runtime cannot authorize production {context} "
+            "(#695). install_test_native_runtime / former allow_test_bootstrap "
+            "must never reach real Gitea mutation endpoints."
+        )
+    assert_sanctioned_mutation_runtime(context)
+
+
+def assert_sanctioned_mutation_runtime(context: str = "mutation") -> None:
+    """Fail closed when mutation code runs outside native MCP transport (#695).
+
+    Under pytest, hermetic unit tests are allowed (profile/permission tests).
+    Outside pytest, requires production-mode transport-bound native runtime.
+    Test-mode records do not authorize non-pytest production mutations.
+    ``GITEA_ALLOW_DIRECT_MCP_IMPORT`` never authorizes mutations (#695 AC1).
+    """
+    if is_pytest_runtime():
+        return
+    # AC1: direct-import env is never a mutation recovery path (PR #701).
+    assert_no_direct_import_bypass(context)
+    if is_production_native_mcp_transport():
+        return
+    mode = (_NATIVE_RUNTIME or {}).get("mode")
+    if mode == _RUNTIME_MODE_TEST:
+        raise UnsanctionedRuntimeError(
+            f"Test-mode native runtime cannot authorize production {context} "
+            "(#695). Test bootstrap cannot reach production mutation endpoints."
+        )
+    env_spoof = (os.environ.get(SANCTIONED_DAEMON_ENV) or "").strip() in {
+        "1",
+        "true",
+        "yes",
+    }
+    extra = ""
+    if env_spoof:
+        extra = (
+            f" Note: {SANCTIONED_DAEMON_ENV} alone is not sufficient (#695); "
+            "native transport requires the official MCP entrypoint and a live "
+            "transport bind."
+        )
+    phase = (_NATIVE_RUNTIME or {}).get("phase")
+    if phase == _PHASE_ENTRYPOINT_CLAIMED:
+        extra = (
+            (extra + " ") if extra else " "
+        ) + (
+            "Entrypoint was claimed but native MCP transport was never bound "
+            "(#695); offline launch/import of the real entrypoint does not "
+            "grant mutation authority."
+        )
+    raise UnsanctionedRuntimeError(
+        f"Unsanctioned / non-native runtime blocked {context} (#695). "
+        "Do not import gitea_mcp_server or call mutation helpers from a raw "
+        "shell, offline runner, or ad-hoc script after native MCP failure. "
+        "Stop and reconnect the official MCP daemon (mcp_server.py) over "
+        "native transport. "
+        f"Do not set {ALLOW_DIRECT_IMPORT_ENV}, override "
+        f"{SESSION_STATE_DIR_ENV}, or use raw token env vars in LLM "
+        f"sessions.{extra}"
+    )
+
+
+def assert_keychain_access_allowed() -> None:
+    """Fail closed for git-credential keychain fill outside sanctioned contexts."""
+    if is_sanctioned_mcp_daemon():
+        return
+    # Operator-only keychain CLI remains available outside LLM mutation path.
+    if (os.environ.get(ALLOW_KEYCHAIN_CLI_ENV) or "").strip() in {"1", "true", "yes"}:
+        if not is_pytest_runtime():
+            # Still block pure env spoof of SANCTIONED_DAEMON for keychain when
+            # FORCE is set for tests.
+            if (os.environ.get(FORCE_PROVENANCE_FAIL_ENV) or "").strip() in {
+                "1",
+                "true",
+                "yes",
+            }:
+                pass
+            else:
+                return
+    raise UnsanctionedRuntimeError(
+        "Unsanctioned keychain/credential fill blocked (#558/#695). "
+        "Token extraction via git-credential is only allowed inside the "
+        f"official native MCP daemon or with explicit operator opt-in "
+        f"{ALLOW_KEYCHAIN_CLI_ENV}=1 (never for offline mutation runners)."
+    )
+
+
+def native_runtime_status() -> dict[str, Any]:
+    """LLM-safe native runtime status (no raw token)."""
+    rt = _NATIVE_RUNTIME or {}
+    return {
+        "native_mcp_transport": is_native_mcp_transport(),
+        "production_native_mcp_transport": is_production_native_mcp_transport(),
+        "pytest": is_pytest_runtime(),
+        "pid": rt.get("pid"),
+        "token_fingerprint": rt.get("token_fingerprint"),
+        "started_at": rt.get("started_at"),
+        "entrypoint": rt.get("entrypoint"),
+        "entrypoint_path": rt.get("entrypoint_path"),
+        "phase": rt.get("phase"),
+        "transport": rt.get("transport"),
+        "mode": rt.get("mode"),
+        "session_state_dir": pinned_session_state_dir() or rt.get("session_state_dir"),
+        "session_state_dir_pinned": pinned_session_state_dir() is not None,
+        "direct_import_env_set": direct_import_env_enabled(),
+        "env_sanctioned_alone_insufficient": True,
+        "sanctioned_env": SANCTIONED_DAEMON_ENV,
+        "allow_direct_import_env": ALLOW_DIRECT_IMPORT_ENV,
+        "session_state_dir_env": SESSION_STATE_DIR_ENV,
+        "allow_keychain_cli_env": ALLOW_KEYCHAIN_CLI_ENV,
+    }
+
+
+def runtime_status() -> dict[str, Any]:
+    """Backward-compatible status payload."""
+    status = native_runtime_status()
+    status["sanctioned_daemon"] = is_sanctioned_mcp_daemon()
+    return status
+
+
+def mutation_provenance_fields() -> dict[str, Any]:
+    """Fields to attach to live mutation / review audit records (#695 AC6)."""
+    st = native_runtime_status()
+    transport = "native_mcp" if st["native_mcp_transport"] else "untrusted"
+    if st.get("mode") == _RUNTIME_MODE_TEST and st["native_mcp_transport"]:
+        transport = "test_native_mcp"
+    return {
+        "transport": transport,
+        "native_mcp_transport": bool(st["native_mcp_transport"]),
+        "production_native_mcp_transport": bool(
+            st.get("production_native_mcp_transport")
+        ),
+        "native_runtime_pid": st.get("pid"),
+        "native_token_fingerprint": st.get("token_fingerprint"),
+        "entrypoint": st.get("entrypoint"),
+        "phase": st.get("phase"),
+        "mode": st.get("mode"),
+        "session_state_dir": st.get("session_state_dir"),
+        "session_state_dir_pinned": bool(st.get("session_state_dir_pinned")),
+    }

mcp_namespace_health.py

@@ -0,0 +1,306 @@
+"""Assess live MCP namespace health without trusting static registration.
+
+The IDE/client namespace can fail with EOF even when this Python process still
+registers the Gitea tools with FastMCP.  These helpers keep that distinction
+explicit so reviewer/merger flows can fail closed on the live path.
+
+Probe sources
+-------------
+* ``client_namespace`` — evidence from the IDE-managed MCP client path (the
+  only source that can prove the workflow namespace is healthy).
+* ``offline_spawn`` — a separate ``subprocess.Popen`` JSON-RPC handshake
+  (e.g. ``test_mcp_conn.py``). Useful offline, but **never** proves the
+  IDE-managed namespace is callable.
+* ``unknown`` — legacy/unspecified; treated as not IDE-proven.
+"""
+
+from __future__ import annotations
+
+from typing import Any
+
+
+REQUIRED_NAMESPACE_TOOLS = {
+    "gitea-author": "gitea_whoami",
+    "gitea-reviewer": "gitea_whoami",
+    "gitea-merger": "gitea_whoami",
+    "gitea-tools": "gitea_list_profiles",
+}
+
+DEFAULT_NAMESPACES = tuple(REQUIRED_NAMESPACE_TOOLS)
+
+# Namespaces that must be healthy for a given mutation task.
+TASK_REQUIRED_NAMESPACES = {
+    "review_pr": "gitea-reviewer",
+    "submit_review": "gitea-reviewer",
+    "merge_pr": "gitea-merger",
+}
+
+PROBE_SOURCE_CLIENT = "client_namespace"
+PROBE_SOURCE_OFFLINE = "offline_spawn"
+PROBE_SOURCE_UNKNOWN = "unknown"
+VALID_PROBE_SOURCES = frozenset(
+    {PROBE_SOURCE_CLIENT, PROBE_SOURCE_OFFLINE, PROBE_SOURCE_UNKNOWN}
+)
+
+EOF_PATTERNS = (
+    "client is closing: eof",
+    "transport closed",
+    "connection closed",
+    "broken pipe",
+    "end of file",
+    "eof",
+)
+
+SAFE_ENV_KEYS = (
+    "GITEA_MCP_PROFILE",
+    "GITEA_PROFILE_NAME",
+    "GITEA_SERVICE",
+    "GITEA_EXECUTION_ROLE",
+    "GITEA_MCP_CONFIG",
+)
+
+
+def _as_list(value: Any) -> list[str] | None:
+    if value is None:
+        return None
+    if isinstance(value, (list, tuple, set)):
+        return [str(v) for v in value]
+    return [str(value)]
+
+
+def _contains_eof(text: str | None) -> bool:
+    lowered = (text or "").lower()
+    return any(pattern in lowered for pattern in EOF_PATTERNS)
+
+
+def _normalize_probe_source(probe_source: str | None) -> str:
+    raw = (probe_source or PROBE_SOURCE_UNKNOWN).strip().lower()
+    if raw in VALID_PROBE_SOURCES:
+        return raw
+    return PROBE_SOURCE_UNKNOWN
+
+
+def _safe_env_summary(process: dict[str, Any] | None) -> dict[str, str]:
+    if not process:
+        return {}
+    env = process.get("env") or process.get("environment") or {}
+    if not isinstance(env, dict):
+        return {}
+    return {
+        key: str(env[key])
+        for key in SAFE_ENV_KEYS
+        if key in env and env[key] not in (None, "")
+    }
+
+
+def classify_namespace_probe(
+    namespace: str,
+    *,
+    required_tool: str | None = None,
+    registered_tools: list[str] | tuple[str, ...] | set[str] | None = None,
+    probe_result: dict[str, Any] | None = None,
+    process: dict[str, Any] | None = None,
+    config_path: str | None = None,
+    profile: str | None = None,
+    configured: bool = True,
+    probe_source: str | None = None,
+) -> dict[str, Any]:
+    """Classify whether a required tool is callable through a live namespace.
+
+    ``registered_tools`` is static/server-side evidence. ``probe_result`` is
+    live invocation evidence. Only ``probe_source=client_namespace`` proves the
+    IDE-managed path; ``offline_spawn`` is an offline subprocess check only.
+    """
+    ns = (namespace or "").strip()
+    tool = required_tool or REQUIRED_NAMESPACE_TOOLS.get(ns) or "gitea_whoami"
+    source = _normalize_probe_source(probe_source)
+    registered_list = _as_list(registered_tools)
+    registered = None if registered_list is None else tool in registered_list
+
+    probe = probe_result or {}
+    probe_success = bool(probe.get("success"))
+    error_message = str(
+        probe.get("error")
+        or probe.get("message")
+        or probe.get("stderr")
+        or probe.get("exception")
+        or ""
+    )
+    error_type = str(probe.get("error_type") or "").strip()
+    if not error_type and error_message:
+        if _contains_eof(error_message):
+            error_type = "namespace_eof"
+        elif "timeout" in error_message.lower():
+            error_type = "namespace_timeout"
+        else:
+            error_type = "namespace_call_failed"
+
+    if not configured:
+        error_type = "namespace_not_configured"
+    elif registered is False:
+        error_type = "tool_missing"
+    elif not probe_result:
+        error_type = "live_probe_missing"
+    elif not probe_success and not error_type:
+        error_type = "namespace_call_failed"
+
+    callable_live = bool(configured and probe_result and probe_success)
+    # Probe-path health (spawn or client). IDE-proven only for client path.
+    healthy = bool(configured and registered is not False and callable_live)
+    ide_namespace_proven = bool(healthy and source == PROBE_SOURCE_CLIENT)
+    process_pid = process.get("pid") if isinstance(process, dict) else None
+    profile_name = profile or (
+        process.get("profile") if isinstance(process, dict) else None
+    )
+    env_summary = _safe_env_summary(process)
+
+    reasons: list[str] = []
+    if not configured:
+        reasons.append(f"MCP namespace '{ns}' is not configured.")
+    if registered is False:
+        reasons.append(
+            f"Required tool '{tool}' is not registered in namespace '{ns}'."
+        )
+    if error_type == "live_probe_missing":
+        reasons.append(
+            f"No live client invocation proof was supplied for '{ns}.{tool}'."
+        )
+    elif error_type == "namespace_eof":
+        reasons.append(
+            f"Live MCP namespace '{ns}' returned EOF while invoking '{tool}'."
+        )
+    elif error_type == "namespace_timeout":
+        reasons.append(
+            f"Live MCP namespace '{ns}' timed out while invoking '{tool}'."
+        )
+    elif error_type == "namespace_call_failed":
+        reasons.append(
+            f"Live MCP namespace '{ns}' failed while invoking '{tool}'."
+        )
+    if source == PROBE_SOURCE_OFFLINE:
+        reasons.append(
+            "Probe source is offline_spawn (subprocess JSON-RPC); this does "
+            "not prove the IDE-managed MCP namespace is healthy."
+        )
+    elif source == PROBE_SOURCE_UNKNOWN and probe_result:
+        reasons.append(
+            "Probe source unspecified; treat as not IDE-namespace proof unless "
+            "re-supplied with probe_source='client_namespace'."
+        )
+
+    remediation = []
+    if not healthy or not ide_namespace_proven:
+        remediation.append(
+            "Reconnect the IDE MCP client namespace (client reconnect / "
+            f"relaunch), then invoke '{tool}' through namespace '{ns}' and "
+            "record the result with probe_source='client_namespace'."
+        )
+        remediation.append(
+            "Do not treat offline subprocess probes (test_mcp_conn.py) or "
+            "shell kill/PID respawn as proof the IDE namespace is repaired."
+        )
+        if process_pid and source != PROBE_SOURCE_OFFLINE:
+            remediation.append(
+                f"Diagnostics may include PID {process_pid}; process details "
+                "are informational only — recovery is client-layer reconnect."
+            )
+    else:
+        remediation.append(
+            f"IDE-managed namespace '{ns}' can invoke '{tool}' "
+            f"(probe_source={source})."
+        )
+
+    # Client-namespace broken health blocks review/merge. Offline probes never
+    # authorize mutations and only block when they report unhealthy (still
+    # fail-closed for known bad spawn evidence).
+    blocks = False
+    if source == PROBE_SOURCE_CLIENT:
+        blocks = namespace_health_blocks_task("merge_pr", healthy)
+    elif source == PROBE_SOURCE_OFFLINE:
+        # Offline never proves IDE health; never unblock. Unhealthy offline
+        # still surfaces as a soft diagnostic, not a mutation-ledger block.
+        blocks = False
+    else:
+        # Unknown source: only block when evidence is unhealthy (fail closed
+        # on bad data without treating success as IDE proof).
+        blocks = namespace_health_blocks_task("merge_pr", healthy)
+
+    return {
+        "success": healthy,
+        "healthy": healthy,
+        "namespace": ns,
+        "required_tool": tool,
+        "configured": configured,
+        "registered_tools_checked": registered_list is not None,
+        "required_tool_registered": registered,
+        "required_tool_callable": callable_live,
+        "probe_source": source,
+        "ide_namespace_proven": ide_namespace_proven,
+        "error_type": None if healthy else error_type,
+        "error_message": error_message or None,
+        "reasons": reasons,
+        "remediation": remediation,
+        "diagnostics": {
+            "namespace": ns,
+            "required_tool": tool,
+            "process_pid": process_pid,
+            "profile": profile_name,
+            "env": env_summary,
+            "config_path": config_path,
+            "probe_source": source,
+        },
+        "blocks_merge_workflow": blocks,
+    }
+
+
+def namespace_health_blocks_task(task: str, healthy: bool) -> bool:
+    """Return whether a broken namespace must block a workflow task."""
+    if healthy:
+        return False
+    return (task or "").strip() in {"merge_pr", "review_pr", "submit_review"}
+
+
+def required_namespace_for_task(task: str) -> str | None:
+    """Map a mutation task to the MCP namespace that must be healthy."""
+    return TASK_REQUIRED_NAMESPACES.get((task or "").strip())
+
+
+def mutation_gate_from_session(
+    task: str,
+    session_health: dict[str, dict[str, Any]] | None,
+) -> list[str]:
+    """Fail-closed gate using recorded client-namespace health assessments.
+
+    * Missing session entry → no gate (caller has not assessed yet).
+    * Client-namespace unhealthy / not IDE-proven → block mutation.
+    * Offline-only session entries never authorize mutations.
+    """
+    ns = required_namespace_for_task(task)
+    if not ns:
+        return []
+    store = session_health or {}
+    entry = store.get(ns)
+    if not entry:
+        return []
+    source = _normalize_probe_source(entry.get("probe_source"))
+    if source != PROBE_SOURCE_CLIENT:
+        return [
+            f"recorded namespace health for '{ns}' is probe_source={source}, "
+            "not client_namespace; re-probe through the IDE-managed path "
+            f"before {(task or 'mutation')}"
+        ]
+    if entry.get("ide_namespace_proven") and entry.get("healthy"):
+        return []
+    if entry.get("blocks_merge_workflow") or not entry.get("healthy"):
+        detail = entry.get("error_type") or "unhealthy"
+        return [
+            f"live MCP namespace '{ns}' is recorded {detail} "
+            f"(probe_source={source}); repair the IDE namespace before "
+            f"{(task or 'mutation')} (fail closed, #543)"
+        ]
+    if not entry.get("ide_namespace_proven"):
+        return [
+            f"live MCP namespace '{ns}' is not IDE-proven; supply a "
+            "client_namespace probe before mutation (fail closed, #543)"
+        ]
+    return []

mcp_native_cleanup_proof.py

@@ -13,6 +13,7 @@ from typing import Any
 
 AUTHORIZED_CLEANUP_TOOLS = frozenset({
     "gitea_cleanup_post_merge_moot_lease",
+    "gitea_cleanup_obsolete_reviewer_comment_lease",
     "gitea_reconcile_merged_cleanups",
     "gitea_delete_branch",
     "gitea_cleanup_merged_pr_branch",

mcp_server.py

@@ -6,6 +6,10 @@ Runs over stdio. All tools authenticate via macOS keychain (git credential fill)
 import os
 import sys
 
+if "PYTEST_CURRENT_TEST" not in os.environ:
+    sys.stderr = open("/tmp/mcp_server_stderr.log", "a", buffering=1)
+sys.stderr.write(f"\n--- MCP SERVER STARTUP (PID {os.getpid()}) ---\n")
+
 from role_session_router import (
     python_bytes_have_conflict_markers,
     skip_python_scan_walk_root,
@@ -37,6 +41,19 @@ def check_conflict_markers():
 
 check_conflict_markers()
 
+# #558 / #695: claim the official entrypoint before loading mutation modules.
+# This alone does NOT authorize mutations — gitea_mcp_server binds the live
+# native MCP transport (stdio) immediately before mcp.run. Import-only or
+# offline launch without that bind fails closed on mutations.
+try:
+    import mcp_daemon_guard
+
+    mcp_daemon_guard.mark_sanctioned_daemon()
+except Exception:
+    # Guard import failures must not hide conflict-marker infra_stop above;
+    # gitea_mcp_server main also marks + binds when run as __main__.
+    pass
+
 # Execute the actual server logic via exec in this namespace.
 impl_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "gitea_mcp_server.py")
 try:

mcp_session_state.py

@@ -30,12 +30,72 @@ DEFAULT_TTL_HOURS = 4.0
 
 KIND_WORKFLOW_LOAD = "review_workflow_load"
 KIND_DECISION_LOCK = "review_decision_lock"
+KIND_REVIEW_DRAFT = "review_draft"
+# Durable marker set when a worker session attempts a direct stable-branch push
+# or a root-checkout local commit (#671). Keyed per profile identity like the
+# other session proofs; a contaminated session fails closed on gated mutations
+# until a reconciler audits and clears it.
+KIND_STABLE_BRANCH_CONTAMINATION = "stable_branch_contamination"
+# #709: archive prior terminal decision ledgers instead of silent overwrite.
+KIND_DECISION_LOCK_ARCHIVE = "review_decision_lock_archive"
+# #709: post-merge cleanup/audit reconciliation-required durable record.
+KIND_POST_MERGE_DECISION_RECOVERY = "post_merge_decision_recovery"
+# #709: truthful record when historical terminal evidence is irrecoverably gone.
+KIND_IRRECOVERABLE_DECISION_PROVENANCE = "irrecoverable_decision_provenance"
+# #709 F1: server-side non-forgeable authorization artifact for recovery.
+KIND_IRRECOVERABLE_PROVENANCE_AUTH = "irrecoverable_provenance_authorization"
+
+# Kinds that must survive the default session-state TTL (forensic / recovery).
+#
+# KIND_DECISION_LOCK is recovery-critical (#720): terminal review provenance is
+# not disposable cache. A generic four-hour TTL must not drop old-head evidence
+# or make ``fresh_review_on_current_head_allowed`` unreachable. Same-head / same-
+# run #332 protections still apply once the ledger is loadable. Other session
+# kinds (workflow load, drafts, etc.) remain TTL-bound.
+RECOVERY_CRITICAL_KINDS = frozenset(
+    {
+        KIND_DECISION_LOCK,
+        KIND_DECISION_LOCK_ARCHIVE,
+        KIND_POST_MERGE_DECISION_RECOVERY,
+        KIND_IRRECOVERABLE_DECISION_PROVENANCE,
+        KIND_IRRECOVERABLE_PROVENANCE_AUTH,
+    }
+)
+
 
 _SAFE_SEGMENT_RE = re.compile(r"[^A-Za-z0-9._+-]+")
 SESSION_PROFILE_LOCK_ENV = "GITEA_SESSION_PROFILE_LOCK"
 
 
 def default_state_dir() -> str:
+    """Resolve the durable session-state root.
+
+    When production native MCP transport is bound (#695 AC2), the directory
+    pinned at transport bind is authoritative: later overrides of
+    ``GITEA_MCP_SESSION_STATE_DIR`` cannot manufacture a second authority
+    domain (PR #701 recurrence: ``.mcp_session_701`` evasion of cross-PR
+    decision locks).
+    """
+    try:
+        import mcp_daemon_guard
+
+        pinned = mcp_daemon_guard.pinned_session_state_dir()
+        if pinned:
+            return pinned
+    except Exception:
+        # Fail open to env/default only when guard is unavailable (e.g. partial
+        # import during bootstrap). Mutation gates still fail closed separately.
+        pass
+    raw = (os.environ.get(STATE_DIR_ENV) or DEFAULT_STATE_DIR).strip()
+    return raw or DEFAULT_STATE_DIR
+
+
+def env_session_state_dir_unpinned() -> str:
+    """Raw env/default session-state dir ignoring production transport pin.
+
+    Intended for diagnostics and tests that assert pin behavior — not for
+    mutation-sensitive durable proofs under a bound native daemon.
+    """
     raw = (os.environ.get(STATE_DIR_ENV) or DEFAULT_STATE_DIR).strip()
     return raw or DEFAULT_STATE_DIR
 
@@ -192,6 +252,16 @@ def _write_json(path: str, data: dict[str, Any]) -> None:
                 pass
 
 
+def is_recovery_critical_record(record: dict[str, Any] | None, kind: str | None = None) -> bool:
+    """True when a durable record must outlive the generic session-state TTL."""
+    if not record and not kind:
+        return False
+    record_kind = ((record or {}).get("kind") or kind or "").strip()
+    if record_kind in RECOVERY_CRITICAL_KINDS:
+        return True
+    return bool((record or {}).get("recovery_critical"))
+
+
 def identity_match_reasons(
     record: dict[str, Any] | None,
     *,
@@ -234,7 +304,9 @@ def identity_match_reasons(
         reasons.append("session state missing recorded_at timestamp (fail closed)")
     else:
         age = _now_utc() - recorded_at
-        if age > timedelta(hours=ttl_hours()):
+        kind = (record.get("kind") or "").strip()
+        ttl_exempt = is_recovery_critical_record(record, kind=kind)
+        if age > timedelta(hours=ttl_hours()) and not ttl_exempt:
             reasons.append(
                 f"session state expired after {ttl_hours():g}h (fail closed)"
             )
@@ -243,6 +315,113 @@ def identity_match_reasons(
     return reasons
 
 
+def inspect_state_envelope(
+    *,
+    kind: str,
+    remote: str | None = None,
+    org: str | None = None,
+    repo: str | None = None,
+    profile_identity: str | None = None,
+    state_dir: str | None = None,
+) -> dict[str, Any]:
+    """Read-only disk inspection for assessment when TTL would otherwise hide state (#720).
+
+    Does **not** apply identity/TTL rejection to the returned presence flags.
+    Callers use this to distinguish:
+      * no file on disk
+      * file present but TTL would reject a non-critical kind
+      * recovery-critical ledger (e.g. KIND_DECISION_LOCK) still loadable
+    Never mutates files. Never returns secrets.
+    """
+    profile = current_profile_identity(profile_identity=profile_identity)
+    path = state_file_path(
+        kind=kind,
+        remote=remote,
+        org=org,
+        repo=repo,
+        profile_identity=profile,
+        state_dir=state_dir,
+    )
+    result: dict[str, Any] = {
+        "kind": kind,
+        "profile_identity": profile,
+        "path_basename": os.path.basename(path),
+        "on_disk": False,
+        "has_payload": False,
+        "recorded_at": None,
+        "updated_at": None,
+        "age_hours": None,
+        "ttl_hours": ttl_hours(),
+        "age_exceeds_default_ttl": False,
+        "recovery_critical": kind in RECOVERY_CRITICAL_KINDS,
+        "ttl_exempt": kind in RECOVERY_CRITICAL_KINDS,
+        "would_ttl_reject": False,
+        "identity_reasons": [],
+        "summary": "no session-state file on disk",
+    }
+    if not path or not os.path.exists(path):
+        return result
+    result["on_disk"] = True
+    envelope = _read_json(path)
+    if not envelope:
+        result["summary"] = "session-state file present but unreadable or empty"
+        return result
+    payload = envelope.get("payload")
+    merged: dict[str, Any] = dict(payload) if isinstance(payload, dict) else {}
+    result["has_payload"] = isinstance(payload, dict)
+    for key in (
+        "kind",
+        "remote",
+        "org",
+        "repo",
+        "profile_identity",
+        "session_profile_lock",
+        "recorded_at",
+        "updated_at",
+        "writer_pid",
+        "recovery_critical",
+    ):
+        if key in envelope and key not in merged:
+            merged[key] = envelope[key]
+    if not merged.get("kind"):
+        merged["kind"] = kind
+    recorded_at = _parse_iso(merged.get("recorded_at") or merged.get("updated_at"))
+    result["recorded_at"] = merged.get("recorded_at") or merged.get("updated_at")
+    result["updated_at"] = merged.get("updated_at") or merged.get("recorded_at")
+    if recorded_at is not None:
+        age = _now_utc() - recorded_at
+        age_hours = age.total_seconds() / 3600.0
+        result["age_hours"] = age_hours
+        result["age_exceeds_default_ttl"] = age > timedelta(hours=ttl_hours())
+    ttl_exempt = is_recovery_critical_record(merged, kind=kind)
+    result["recovery_critical"] = ttl_exempt
+    result["ttl_exempt"] = ttl_exempt
+    identity_reasons = identity_match_reasons(
+        merged,
+        remote=remote,
+        org=org,
+        repo=repo,
+        profile_identity=profile,
+    )
+    result["identity_reasons"] = list(identity_reasons)
+    result["would_ttl_reject"] = any("expired" in r for r in identity_reasons)
+    if result["has_payload"] and ttl_exempt:
+        result["summary"] = (
+            "recovery-critical session-state present on disk and TTL-exempt; "
+            "load via load_state for full payload"
+        )
+    elif result["has_payload"] and result["would_ttl_reject"]:
+        result["summary"] = (
+            "session-state file present on disk but generic TTL would reject load "
+            f"(age_hours={result.get('age_hours')!r}, ttl={ttl_hours():g}h)"
+        )
+    elif result["has_payload"]:
+        result["summary"] = "session-state file present and within TTL / identity gates"
+    else:
+        result["summary"] = "session-state file present without a dict payload"
+    return result
+
+
 def load_state(
     *,
     kind: str,
@@ -354,6 +533,26 @@ def save_state(
             body["org"] = key_org
         if key_repo is not None:
             body["repo"] = key_repo
+        # Stamp session-state authority used for this write (#695 AC2 / AC6).
+        body.setdefault("session_state_dir", root)
+        try:
+            import mcp_daemon_guard
+
+            prov = mcp_daemon_guard.mutation_provenance_fields()
+            body.setdefault(
+                "native_token_fingerprint", prov.get("native_token_fingerprint")
+            )
+            body.setdefault(
+                "native_mcp_transport", bool(prov.get("native_mcp_transport"))
+            )
+            body.setdefault(
+                "production_native_mcp_transport",
+                bool(prov.get("production_native_mcp_transport")),
+            )
+            body.setdefault("transport", prov.get("transport"))
+        except Exception:
+            body.setdefault("native_mcp_transport", False)
+            body.setdefault("transport", "untrusted")
 
         envelope = {
             "kind": kind,
@@ -365,6 +564,8 @@ def save_state(
             "recorded_at": body["recorded_at"],
             "updated_at": body["updated_at"],
             "writer_pid": body["writer_pid"],
+            "session_state_dir": body.get("session_state_dir"),
+            "transport": body.get("transport"),
             "payload": body,
         }
         _write_json(path, envelope)
@@ -389,3 +590,177 @@ def clear_state(
         profile_identity=profile_identity,
         state_dir=state_dir,
     )
+
+def list_decision_lock_profile_identities(
+    state_dir: str | None = None,
+) -> list[str]:
+    """Return profile identities that have a durable review_decision_lock file (#709).
+
+    Filename form: ``review_decision_lock-<profile>.json`` (see ``state_key``).
+    Does not validate TTL or identity — callers must load via ``load_state``.
+    """
+    root = (state_dir or default_state_dir()).strip()
+    if not root or not os.path.isdir(root):
+        return []
+    prefix = f"{_sanitize_segment(KIND_DECISION_LOCK)}-"
+    suffix = ".json"
+    found: list[str] = []
+    try:
+        names = os.listdir(root)
+    except OSError:
+        return []
+    for name in names:
+        if not name.startswith(prefix) or not name.endswith(suffix):
+            continue
+        if name.endswith(".lock"):
+            continue
+        mid = name[len(prefix) : -len(suffix)]
+        if mid:
+            found.append(mid)
+    return sorted(set(found))
+
+
+def load_state_for_profile(
+    *,
+    kind: str,
+    profile_identity: str,
+    remote: str | None = None,
+    org: str | None = None,
+    repo: str | None = None,
+    state_dir: str | None = None,
+    skip_identity_match: bool = False,
+    enforce_repo_scope: bool = True,
+) -> dict[str, Any] | None:
+    """Load durable state for an explicit profile identity (#709 cross-profile).
+
+    When *skip_identity_match* is True, still requires the file's recorded
+    profile_identity to equal the requested profile (anti-stomp), but does not
+    require the *active* session identity to match — needed so a merger can
+    inspect a reviewer lock after merge.
+
+    #709 F3: remote/org/repo filter reasons are **enforced** (not merely
+    computed). When *enforce_repo_scope* is True (default) and the caller
+    supplies remote/org/repo, mismatches fail closed with ``None``.
+    """
+    # #709 F3: refuse traversal / malformed profile identities.
+    try:
+        from irrecoverable_provenance import assess_profile_path_identity
+
+        path_gate = assess_profile_path_identity(profile_identity)
+        if not path_gate.get("valid"):
+            return None
+    except Exception:
+        # Module may be mid-import in edge bootstraps; fall through to
+        # conservative checks below.
+        raw = str(profile_identity or "")
+        if ".." in raw or "/" in raw or "\\" in raw or "\x00" in raw:
+            return None
+
+    profile = current_profile_identity(profile_identity=profile_identity)
+    root = _ensure_state_dir(state_dir)
+    # Refuse symlink escape of the state root (#709 F3).
+    try:
+        real_root = os.path.realpath(root)
+        path = state_file_path(
+            kind=kind,
+            remote=remote,
+            org=org,
+            repo=repo,
+            profile_identity=profile,
+            state_dir=root,
+        )
+        real_path = os.path.realpath(path) if os.path.exists(path) else path
+        if os.path.exists(path) and not str(real_path).startswith(
+            str(real_root) + os.sep
+        ) and str(real_path) != str(real_root):
+            return None
+    except OSError:
+        return None
+
+    path = state_file_path(
+        kind=kind,
+        remote=remote,
+        org=org,
+        repo=repo,
+        profile_identity=profile,
+        state_dir=root,
+    )
+    lock_path = f"{path}.lock"
+    with _exclusive_file_lock(lock_path):
+        envelope = _read_json(path)
+    if not envelope:
+        return None
+    payload = envelope.get("payload")
+    if not isinstance(payload, dict):
+        return None
+    merged = dict(payload)
+    for key in (
+        "kind",
+        "remote",
+        "org",
+        "repo",
+        "profile_identity",
+        "session_profile_lock",
+        "recorded_at",
+        "updated_at",
+        "writer_pid",
+    ):
+        if key in envelope and key not in merged:
+            merged[key] = envelope[key]
+    stored = (merged.get("profile_identity") or "").strip()
+    if stored and stored != profile:
+        return None
+    if skip_identity_match:
+        # Enforce remote/org/repo when caller provides them (#709 F3).
+        # Drop only *active session* profile-identity mismatches; keep
+        # expiry, spoof, and repository-scope reasons.
+        scope_remote = remote if enforce_repo_scope else None
+        scope_org = org if enforce_repo_scope else None
+        scope_repo = repo if enforce_repo_scope else None
+        reasons = identity_match_reasons(
+            merged,
+            remote=scope_remote,
+            org=scope_org,
+            repo=scope_repo,
+            profile_identity=stored or profile,
+        )
+        filtered = [
+            r
+            for r in reasons
+            if "profile identity mismatch" not in r
+            or (stored and stored != profile)
+        ]
+        # When caller requested a specific remote/org/repo, also fail if the
+        # stored record lacks those identity fields entirely (legacy incomplete).
+        if enforce_repo_scope:
+            for field, want in (
+                ("remote", remote),
+                ("org", org),
+                ("repo", repo),
+            ):
+                want_s = (want or "").strip()
+                if not want_s:
+                    continue
+                have = (str(merged.get(field) or "")).strip()
+                if not have:
+                    filtered.append(
+                        f"session state {field} missing on durable record "
+                        f"(expected={want_s!r}; fail closed, #709 F3)"
+                    )
+                elif have != want_s:
+                    # identity_match_reasons already adds mismatch; ensure kept
+                    pass
+        if filtered:
+            return None
+        return merged
+    reasons = identity_match_reasons(
+        merged,
+        remote=remote,
+        org=org,
+        repo=repo,
+        profile_identity=profile,
+    )
+    if reasons:
+        return None
+    return merged
+

mcp_tool_error_boundary.py

@@ -0,0 +1,451 @@
+"""MCP tool-boundary error mapping for known Gitea client failures (#699).
+
+Known authentication / authorization / network / configuration failures leave
+the tool boundary as a sanitized structured ``CallToolResult`` with
+``isError=True``. Stdio transport remains connected.
+
+Design constraints (reviewer-ratified, #699 / PR #701):
+
+1. **No secret material** in tool results or daemon logs. Messages are fixed
+   constants keyed by ``reason_code``; HTTP bodies / exception text never
+   surface. Sanitization failure fails closed to ``internal_error``.
+2. **Narrow boundary** wraps the original FastMCP ``Tool.run`` success path;
+   re-raises framework control-flow exceptions (``UrlElicitationRequiredError``);
+   maps only typed client failures. Installation is idempotent.
+3. **No RuntimeError substring heuristics.** Only explicit typed
+   authentication / authorization / network / configuration exceptions
+   receive those labels.
+4. **HTTP classification** lives in ``gitea_auth.classify_http_status``;
+   every 403 is authorization-class.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import sys
+from typing import Any
+
+logger = logging.getLogger("gitea_mcp.tool_error_boundary")
+
+# Stable reason codes (#699).
+REASON_AUTH_FAILED = "auth_failed"
+REASON_AUTH_INVALID_TOKEN = "auth_invalid_token"
+REASON_AUTHZ_INSUFFICIENT_SCOPE = "authz_insufficient_scope"
+REASON_AUTHZ_DENIED = "authz_denied"
+REASON_NETWORK_ERROR = "network_error"
+REASON_CONFIG_ERROR = "config_error"
+REASON_INTERNAL_ERROR = "internal_error"
+REASON_UPSTREAM_UNAVAILABLE = "upstream_unavailable"
+REASON_HTTP_ERROR = "http_error"
+
+ERROR_CLASS_AUTHENTICATION = "authentication"
+ERROR_CLASS_AUTHORIZATION = "authorization"
+ERROR_CLASS_NETWORK = "network"
+ERROR_CLASS_CONFIGURATION = "configuration"
+ERROR_CLASS_INTERNAL = "internal"
+ERROR_CLASS_UPSTREAM = "upstream"
+
+# Fixed, secret-free operator messages. Never interpolate HTTP bodies,
+# Keychain contents, tokens, or arbitrary exception text.
+FIXED_MESSAGES: dict[str, str] = {
+    REASON_AUTH_FAILED: "Gitea authentication failed",
+    REASON_AUTH_INVALID_TOKEN: (
+        "Gitea authentication failed: invalid or revoked credentials"
+    ),
+    REASON_AUTHZ_INSUFFICIENT_SCOPE: (
+        "Gitea authorization failed: insufficient token scope"
+    ),
+    REASON_AUTHZ_DENIED: "Gitea authorization failed: access denied",
+    REASON_NETWORK_ERROR: "Network error contacting Gitea",
+    REASON_CONFIG_ERROR: "Gitea configuration or credential resolution failed",
+    REASON_INTERNAL_ERROR: "Internal tool error",
+    REASON_UPSTREAM_UNAVAILABLE: "Gitea upstream unavailable",
+    REASON_HTTP_ERROR: "Gitea HTTP request failed",
+}
+
+_INSTALL_FLAG = "_gitea_auth_boundary_installed"
+_ORIGINAL_ATTR = "_gitea_auth_boundary_original"
+
+
+def fixed_message(reason_code: str) -> str:
+    """Return the fixed sanitized message for *reason_code* (fail closed)."""
+    return FIXED_MESSAGES.get(reason_code, FIXED_MESSAGES[REASON_INTERNAL_ERROR])
+
+
+def _safe_profile_name() -> str | None:
+    try:
+        from gitea_auth import get_profile
+
+        name = (get_profile() or {}).get("profile_name")
+        if name is None:
+            return None
+        text = str(name).strip()
+        # Profile names are non-secret identifiers; still bound length.
+        return text[:80] if text else None
+    except Exception:
+        return None
+
+
+def _is_framework_control_flow(exc: BaseException) -> bool:
+    """True for exceptions the MCP framework must re-raise unchanged."""
+    try:
+        from mcp.shared.exceptions import UrlElicitationRequiredError
+
+        if isinstance(exc, UrlElicitationRequiredError):
+            return True
+    except Exception:
+        pass
+    # BaseException subclasses that must never become tool isError payloads.
+    if isinstance(exc, (KeyboardInterrupt, SystemExit, GeneratorExit)):
+        return True
+    return False
+
+
+def is_known_client_failure(exc: BaseException) -> bool:
+    """True only for explicit typed Gitea client / config failures.
+
+    Never true for arbitrary ``RuntimeError`` (reviewer finding #3).
+    """
+    try:
+        import gitea_auth
+
+        if isinstance(
+            exc,
+            (
+                gitea_auth.GiteaAuthError,
+                gitea_auth.GiteaAuthzError,
+                gitea_auth.GiteaNetworkError,
+                gitea_auth.GiteaConfigError,
+                gitea_auth.GiteaHttpError,
+            ),
+        ):
+            return True
+    except Exception:
+        return False
+    try:
+        import gitea_config
+
+        if isinstance(exc, gitea_config.ConfigError):
+            return True
+    except Exception:
+        pass
+    return False
+
+
+def classify_exception(exc: BaseException) -> dict[str, Any]:
+    """Return structured classification with **fixed** messages only.
+
+    Only typed authentication / authorization / network / configuration
+    failures receive those labels. Unexpected programming failures are
+    ``internal_error``. HTTP bodies and exception text are never copied
+    into ``message``.
+    """
+    try:
+        return _classify_exception_impl(exc)
+    except Exception:
+        # Fail closed: sanitization / classification failure must not leak.
+        return {
+            "reason_code": REASON_INTERNAL_ERROR,
+            "error_class": ERROR_CLASS_INTERNAL,
+            "http_status": None,
+            "message": fixed_message(REASON_INTERNAL_ERROR),
+            "transport_survives": True,
+        }
+
+
+def _classify_exception_impl(exc: BaseException) -> dict[str, Any]:
+    import gitea_auth
+
+    if isinstance(exc, gitea_auth.GiteaAuthError):
+        code = getattr(exc, "reason_code", None) or REASON_AUTH_INVALID_TOKEN
+        if code not in (
+            REASON_AUTH_FAILED,
+            REASON_AUTH_INVALID_TOKEN,
+        ):
+            code = REASON_AUTH_INVALID_TOKEN
+        return {
+            "reason_code": code,
+            "error_class": ERROR_CLASS_AUTHENTICATION,
+            "http_status": getattr(exc, "http_status", None) or 401,
+            "message": fixed_message(code),
+            "transport_survives": True,
+        }
+    if isinstance(exc, gitea_auth.GiteaAuthzError):
+        code = getattr(exc, "reason_code", None) or REASON_AUTHZ_DENIED
+        if code not in (REASON_AUTHZ_DENIED, REASON_AUTHZ_INSUFFICIENT_SCOPE):
+            code = REASON_AUTHZ_DENIED
+        return {
+            "reason_code": code,
+            "error_class": ERROR_CLASS_AUTHORIZATION,
+            "http_status": getattr(exc, "http_status", None) or 403,
+            "message": fixed_message(code),
+            "transport_survives": True,
+        }
+    if isinstance(exc, gitea_auth.GiteaNetworkError):
+        return {
+            "reason_code": REASON_NETWORK_ERROR,
+            "error_class": ERROR_CLASS_NETWORK,
+            "http_status": getattr(exc, "http_status", None),
+            "message": fixed_message(REASON_NETWORK_ERROR),
+            "transport_survives": True,
+        }
+    if isinstance(exc, gitea_auth.GiteaConfigError):
+        return {
+            "reason_code": REASON_CONFIG_ERROR,
+            "error_class": ERROR_CLASS_CONFIGURATION,
+            "http_status": getattr(exc, "http_status", None),
+            "message": fixed_message(REASON_CONFIG_ERROR),
+            "transport_survives": True,
+        }
+    if isinstance(exc, gitea_auth.GiteaHttpError):
+        code = getattr(exc, "reason_code", None) or REASON_HTTP_ERROR
+        if code == REASON_UPSTREAM_UNAVAILABLE:
+            error_class = ERROR_CLASS_UPSTREAM
+        else:
+            error_class = ERROR_CLASS_INTERNAL
+            code = REASON_HTTP_ERROR
+        return {
+            "reason_code": code,
+            "error_class": error_class,
+            "http_status": getattr(exc, "http_status", None),
+            "message": fixed_message(code),
+            "transport_survives": True,
+        }
+
+    try:
+        import gitea_config
+
+        if isinstance(exc, gitea_config.ConfigError):
+            return {
+                "reason_code": REASON_CONFIG_ERROR,
+                "error_class": ERROR_CLASS_CONFIGURATION,
+                "http_status": None,
+                "message": fixed_message(REASON_CONFIG_ERROR),
+                "transport_survives": True,
+            }
+    except Exception:
+        pass
+
+    # Unwrap FastMCP ToolError cause when the original was a typed failure.
+    cause = getattr(exc, "__cause__", None)
+    if cause is not None and cause is not exc and is_known_client_failure(cause):
+        return _classify_exception_impl(cause)
+
+    # No message-substring authentication heuristics (reviewer finding #3).
+    return {
+        "reason_code": REASON_INTERNAL_ERROR,
+        "error_class": ERROR_CLASS_INTERNAL,
+        "http_status": None,
+        "message": fixed_message(REASON_INTERNAL_ERROR),
+        "transport_survives": True,
+    }
+
+
+def build_structured_error_payload(
+    classification: dict[str, Any],
+    *,
+    tool_name: str | None = None,
+    profile_name: str | None = None,
+) -> dict[str, Any]:
+    """LLM-safe structured payload — fixed message + typed metadata only."""
+    try:
+        reason = str(classification.get("reason_code") or REASON_INTERNAL_ERROR)
+        message = fixed_message(reason)
+        # Refuse to emit any classification message that is not the fixed constant.
+        if classification.get("message") != message:
+            message = fixed_message(reason)
+        payload: dict[str, Any] = {
+            "success": False,
+            "isError": True,
+            "reason_code": reason if reason in FIXED_MESSAGES else REASON_INTERNAL_ERROR,
+            "error_class": classification.get("error_class") or ERROR_CLASS_INTERNAL,
+            "message": message,
+            "transport_survives": True,
+            "retryable": classification.get("error_class")
+            in {
+                ERROR_CLASS_AUTHENTICATION,
+                ERROR_CLASS_NETWORK,
+                ERROR_CLASS_CONFIGURATION,
+            },
+        }
+        status = classification.get("http_status")
+        if isinstance(status, int):
+            payload["http_status"] = status
+        if tool_name and isinstance(tool_name, str):
+            # Tool names are identifiers, not secrets; bound length.
+            payload["tool"] = tool_name[:120]
+        if profile_name and isinstance(profile_name, str):
+            payload["profile"] = profile_name[:80]
+        return payload
+    except Exception:
+        return {
+            "success": False,
+            "isError": True,
+            "reason_code": REASON_INTERNAL_ERROR,
+            "error_class": ERROR_CLASS_INTERNAL,
+            "message": fixed_message(REASON_INTERNAL_ERROR),
+            "transport_survives": True,
+            "retryable": False,
+        }
+
+
+def log_sanitized_daemon_reason(
+    classification: dict[str, Any],
+    *,
+    tool_name: str | None = None,
+    stream=None,
+) -> None:
+    """Daemon log: reason codes only — never exception text or response bodies."""
+    stream = stream if stream is not None else sys.stderr
+    try:
+        reason = classification.get("reason_code") or REASON_INTERNAL_ERROR
+        error_class = classification.get("error_class") or ERROR_CLASS_INTERNAL
+        # Only emit known tokens; never classification['message'] from callers
+        # that might have been poisoned.
+        if reason not in FIXED_MESSAGES:
+            reason = REASON_INTERNAL_ERROR
+            error_class = ERROR_CLASS_INTERNAL
+        parts = [
+            "mcp_tool_error",
+            f"reason_code={reason}",
+            f"error_class={error_class}",
+        ]
+        if tool_name and isinstance(tool_name, str):
+            parts.append(f"tool={tool_name[:120]}")
+        status = classification.get("http_status")
+        if isinstance(status, int):
+            parts.append(f"http_status={status}")
+        # Intentionally no detail= / message= field — secrets lived there.
+        line = " ".join(parts)
+        stream.write(line + "\n")
+        if hasattr(stream, "flush"):
+            stream.flush()
+        logger.warning(line)
+    except Exception:
+        # Fail closed: never fall back to logging the exception.
+        try:
+            stream.write(
+                "mcp_tool_error reason_code=internal_error "
+                "error_class=internal\n"
+            )
+            if hasattr(stream, "flush"):
+                stream.flush()
+        except Exception:
+            pass
+
+
+def to_call_tool_result(
+    exc: BaseException,
+    *,
+    tool_name: str | None = None,
+    profile_name: str | None = None,
+    log: bool = True,
+) -> Any:
+    """Build a FastMCP ``CallToolResult`` with ``isError=True`` for *exc*."""
+    from mcp.types import CallToolResult, TextContent
+
+    try:
+        classification = classify_exception(exc)
+        if log:
+            log_sanitized_daemon_reason(classification, tool_name=tool_name)
+        payload = build_structured_error_payload(
+            classification, tool_name=tool_name, profile_name=profile_name
+        )
+        text = json.dumps(payload, indent=2, sort_keys=True)
+        return CallToolResult(
+            content=[TextContent(type="text", text=text)],
+            structuredContent=payload,
+            isError=True,
+        )
+    except Exception:
+        # Absolute fail-closed path — no exception text.
+        fallback = {
+            "success": False,
+            "isError": True,
+            "reason_code": REASON_INTERNAL_ERROR,
+            "error_class": ERROR_CLASS_INTERNAL,
+            "message": fixed_message(REASON_INTERNAL_ERROR),
+            "transport_survives": True,
+            "retryable": False,
+        }
+        return CallToolResult(
+            content=[
+                TextContent(
+                    type="text",
+                    text=json.dumps(fallback, indent=2, sort_keys=True),
+                )
+            ],
+            structuredContent=fallback,
+            isError=True,
+        )
+
+
+def install_tool_run_boundary(Tool) -> bool:
+    """Install a **narrow** error boundary around FastMCP ``Tool.run``.
+
+    Strategy (preserves framework semantics):
+
+    * Call the **original** ``Tool.run`` for the success path (async, return
+      types, convert_result, protocol behavior unchanged).
+    * Re-raise ``UrlElicitationRequiredError`` and other control-flow
+      exceptions without mapping to ``internal_error``.
+    * Map typed Gitea client failures (and ToolError whose ``__cause__`` is
+      typed) to structured ``CallToolResult(isError=True)``.
+    * Map remaining unexpected tool failures to fixed ``internal_error``
+      isError results (transport survival) without secret-bearing text.
+    * Idempotent: second install is a no-op and returns ``False``.
+    """
+    if getattr(Tool.run, _INSTALL_FLAG, False):
+        return False
+
+    from mcp.server.fastmcp.exceptions import ToolError
+    from mcp.shared.exceptions import UrlElicitationRequiredError
+
+    original_run = Tool.run
+
+    async def run_boundary(
+        self,
+        arguments: dict[str, Any],
+        context=None,
+        convert_result: bool = False,
+    ) -> Any:
+        try:
+            return await original_run(
+                self,
+                arguments,
+                context=context,
+                convert_result=convert_result,
+            )
+        except UrlElicitationRequiredError:
+            # Framework control-flow — must not become internal_error.
+            raise
+        except BaseException as exc:
+            if _is_framework_control_flow(exc):
+                raise
+            if not isinstance(exc, Exception):
+                raise
+
+            # Prefer typed cause under FastMCP ToolError wrappers.
+            target: BaseException = exc
+            if isinstance(exc, ToolError) and exc.__cause__ is not None:
+                target = exc.__cause__
+
+            # Known client failures → structured isError with reason codes.
+            # Unexpected failures → fixed internal_error isError (no secrets).
+            profile_name = _safe_profile_name()
+            return to_call_tool_result(
+                target,
+                tool_name=getattr(self, "name", None),
+                profile_name=profile_name,
+            )
+
+    setattr(run_boundary, _INSTALL_FLAG, True)
+    setattr(run_boundary, _ORIGINAL_ATTR, original_run)
+    Tool.run = run_boundary  # type: ignore[method-assign]
+    return True
+
+
+def boundary_is_installed(Tool) -> bool:
+    """True when the #699 boundary is active on *Tool.run*."""
+    return bool(getattr(Tool.run, _INSTALL_FLAG, False))

merge_approval_gate.py

@@ -1,7 +1,8 @@
-"""Merge approval must pin the current PR head SHA (#471).
+"""Merge approval must pin the current PR head SHA (#471 / #695).
 
 Formal APPROVED reviews that predate the live PR head must not satisfy
-``gitea_merge_pr`` eligibility. Pure assessment helpers are isolated here
+``gitea_merge_pr`` eligibility. Contaminated / quarantined approvals (#695)
+must not authorize merge either. Pure assessment helpers are isolated here
 for hermetic unit tests apart from MCP HTTP calls.
 """
 
@@ -12,6 +13,7 @@ def assess_merge_approval_head(
     *,
     current_head_sha: str | None,
     latest_by_reviewer: dict,
+    quarantined_review_ids: set | None = None,
 ) -> dict:
     """Return whether a visible approval applies to the live PR head.
 
@@ -19,18 +21,43 @@ def assess_merge_approval_head(
         current_head_sha: Current PR head commit SHA.
         latest_by_reviewer: Map of reviewer login → review entry dicts with
             ``verdict``, ``dismissed``, and ``reviewed_head_sha`` keys.
+            Optional ``review_id`` / ``id`` used for quarantine checks (#695).
+        quarantined_review_ids: Optional set of review IDs that must not count
+            toward merge authorization (#695).
 
     Returns:
         dict with ``approval_at_current_head``, ``latest_approved_head_sha``,
         and ``stale_approval_block_reason`` (set when merge must fail closed).
     """
     current = (current_head_sha or "").strip()
-    approved_entries = [
-        entry
-        for entry in (latest_by_reviewer or {}).values()
-        if (entry.get("verdict") or "").upper() == "APPROVED"
-        and not entry.get("dismissed")
-    ]
+    blocked_ids = {int(x) for x in (quarantined_review_ids or set()) if x is not None}
+
+    def _rid(entry: dict):
+        raw = entry.get("review_id", entry.get("id"))
+        try:
+            return int(raw) if raw is not None else None
+        except (TypeError, ValueError):
+            return None
+
+    approved_entries = []
+    quarantined_at_head = []
+    for entry in (latest_by_reviewer or {}).values():
+        if (entry.get("verdict") or "").upper() != "APPROVED":
+            continue
+        if entry.get("dismissed"):
+            continue
+        rid = _rid(entry)
+        head = (entry.get("reviewed_head_sha") or "").strip()
+        if rid is not None and rid in blocked_ids:
+            if current and head == current:
+                quarantined_at_head.append(entry)
+            continue
+        if entry.get("quarantined"):
+            if current and head == current:
+                quarantined_at_head.append(entry)
+            continue
+        approved_entries.append(entry)
+
     at_current = any(
         (entry.get("reviewed_head_sha") or "").strip() == current
         for entry in approved_entries
@@ -47,7 +74,13 @@ def assess_merge_approval_head(
         )[-1]
         latest_approved = (latest_entry.get("reviewed_head_sha") or "").strip() or None
     reason = None
-    if approved_entries and not at_current:
+    if quarantined_at_head and not at_current:
+        reason = (
+            "contaminated/quarantined approval at current head is void for "
+            "merge authorization (#695); required next action: fresh native "
+            "MCP re-review after controller quarantine evidence is recorded"
+        )
+    elif approved_entries and not at_current:
         reason = (
             f"stale approval: approved SHA '{latest_approved}' does not match "
             f"current live PR head SHA '{current or '(unknown)'}' (fail closed); "
@@ -58,4 +91,5 @@ def assess_merge_approval_head(
         "approval_at_current_head": at_current,
         "latest_approved_head_sha": latest_approved,
         "stale_approval_block_reason": reason,
+        "quarantined_approvals_at_current_head": len(quarantined_at_head),
     }

merged_cleanup_reconcile.py

@@ -17,6 +17,7 @@ import issue_lock_store
 
 PROTECTED_BRANCHES = frozenset({"master", "main", "dev"})
 CLOSES_FIXES_RE = re.compile(r"\b(?:closes|fixes)\s+#(\d+)\b", re.IGNORECASE)
+ISSUE_MARKER_RE = re.compile(r"(?:^|[-_/])issue-(\d+)(?:[-_/]|$)", re.IGNORECASE)
 # Reviewer scratch folders: review-pr<N> or review-pr<N>-submit (#534).
 REVIEWER_SCRATCH_NAME_RE = re.compile(
     r"^review-pr(?P<pr>\d+)(?:-submit)?$",
@@ -41,6 +42,59 @@ def resolve_worktree_path(project_root: str, branch: str) -> str:
     return os.path.join(project_root, "branches", branch_worktree_folder(branch))
 
 
+def extract_issue_marker(*values: str | None) -> int | None:
+    """Return the first issue marker found in branch/path-like values."""
+    for value in values:
+        match = ISSUE_MARKER_RE.search(value or "")
+        if match:
+            return int(match.group(1))
+    return None
+
+
+def list_local_worktrees(project_root: str) -> list[dict[str, Any]]:
+    """Parse ``git worktree list --porcelain`` into structured entries."""
+    res = subprocess.run(
+        ["git", "-C", project_root, "worktree", "list", "--porcelain"],
+        capture_output=True,
+        text=True,
+        check=False,
+    )
+    if res.returncode != 0:
+        return []
+
+    entries: list[dict[str, Any]] = []
+    current: dict[str, Any] | None = None
+
+    def flush() -> None:
+        nonlocal current
+        if current:
+            entries.append(current)
+        current = None
+
+    for raw_line in (res.stdout or "").splitlines():
+        line = raw_line.strip()
+        if not line:
+            flush()
+            continue
+        if line.startswith("worktree "):
+            flush()
+            current = {"path": line[9:].strip()}
+        elif current is not None and line.startswith("HEAD "):
+            current["head_sha"] = line[5:].strip()
+        elif current is not None and line.startswith("branch "):
+            ref = line[7:].strip()
+            current["branch_ref"] = ref
+            current["branch"] = (
+                ref[len("refs/heads/"):]
+                if ref.startswith("refs/heads/")
+                else ref
+            )
+        elif current is not None and line == "detached":
+            current["detached"] = True
+    flush()
+    return entries
+
+
 def _git_worktree_porcelain(project_root: str) -> list[dict[str, str | None]]:
     """Parse ``git worktree list --porcelain`` into structured records."""
     res = subprocess.run(
@@ -82,14 +136,115 @@ def _git_worktree_porcelain(project_root: str) -> list[dict[str, str | None]]:
 def discover_local_worktrees(project_root: str) -> dict[str, str]:
     """Return a mapping from branch name to absolute worktree path (#528)."""
     mapping: dict[str, str] = {}
-    for record in _git_worktree_porcelain(project_root):
-        branch_name = record.get("branch")
-        path = record.get("worktree")
-        if branch_name and path:
-            mapping[str(branch_name)] = str(path)
+    for entry in list_local_worktrees(project_root):
+        branch = entry.get("branch")
+        path = entry.get("path")
+        if branch and path:
+            mapping[str(branch)] = str(path)
     return mapping
 
 
+def _is_under_branches(project_root: str, path: str) -> bool:
+    branches_root = os.path.realpath(os.path.join(project_root, "branches"))
+    real_path = os.path.realpath(path)
+    return real_path == branches_root or real_path.startswith(branches_root + os.sep)
+
+
+def _head_is_safe_for_cleanup(
+    *,
+    project_root: str,
+    candidate_head: str | None,
+    pr_head_sha: str | None,
+    target_ref: str | None,
+) -> bool:
+    if not candidate_head:
+        return False
+    if pr_head_sha and candidate_head == pr_head_sha:
+        return True
+    if target_ref:
+        return is_head_ancestor_of_ref(project_root, candidate_head, target_ref) is True
+    return False
+
+
+def resolve_cleanup_worktree_state(
+    *,
+    project_root: str,
+    head_branch: str,
+    issue_number: int | None,
+    pr_head_sha: str | None = None,
+    target_ref: str | None = None,
+) -> dict[str, Any]:
+    """Find the exact or safe alias worktree used for local cleanup (#532)."""
+    expected_path = resolve_worktree_path(project_root, head_branch)
+    state = read_local_worktree_state(expected_path)
+    state["worktree_path"] = expected_path
+    state["expected_worktree_path"] = expected_path
+    state["discovered_worktree_path"] = expected_path if state.get("exists") else None
+    state["match_type"] = "derived_path" if state.get("exists") else "none"
+    state["candidate_paths"] = []
+    state["alias_block_reasons"] = []
+    if state.get("exists"):
+        return state
+
+    branch_issue = extract_issue_marker(head_branch)
+    wanted_issue = issue_number or branch_issue
+    candidates: list[dict[str, Any]] = []
+    unsafe_matches: list[str] = []
+
+    for entry in list_local_worktrees(project_root):
+        path = entry.get("path") or ""
+        if not path or not _is_under_branches(project_root, path):
+            continue
+        branch = entry.get("branch") or ""
+        path_issue = extract_issue_marker(os.path.basename(path), path)
+        entry_issue = extract_issue_marker(branch) or path_issue
+        branch_match = branch == head_branch
+        issue_match = wanted_issue is not None and entry_issue == wanted_issue
+        if not (branch_match or issue_match):
+            continue
+        safe_head = _head_is_safe_for_cleanup(
+            project_root=project_root,
+            candidate_head=entry.get("head_sha"),
+            pr_head_sha=pr_head_sha,
+            target_ref=target_ref,
+        )
+        if not safe_head and branch_match and not pr_head_sha and not target_ref:
+            safe_head = True
+        if not safe_head:
+            unsafe_matches.append(path)
+            continue
+        candidates.append(entry)
+
+    state["candidate_paths"] = [entry.get("path") for entry in candidates]
+    if len(candidates) == 1:
+        path = candidates[0]["path"]
+        alias_state = read_local_worktree_state(path)
+        alias_state["worktree_path"] = path
+        alias_state["expected_worktree_path"] = expected_path
+        alias_state["discovered_worktree_path"] = path
+        alias_state["match_type"] = (
+            "branch" if candidates[0].get("branch") == head_branch else "alias"
+        )
+        alias_state["candidate_paths"] = [path]
+        alias_state["alias_block_reasons"] = []
+        alias_state["alias_verified"] = True
+        return alias_state
+    if len(candidates) > 1:
+        state["alias_block_reasons"].append(
+            "ambiguous local worktree aliases: " + ", ".join(
+                sorted(str(entry.get("path")) for entry in candidates)
+            )
+        )
+        state["match_type"] = "ambiguous_alias"
+    elif unsafe_matches:
+        state["alias_block_reasons"].append(
+            "matching local worktree aliases did not point to the PR head or target branch: "
+            + ", ".join(sorted(unsafe_matches))
+        )
+        state["match_type"] = "unsafe_alias"
+    return state
+
+
 def parse_reviewer_scratch_folder(folder_name: str) -> int | None:
     """Return PR number for a reviewer scratch folder name, else None (#534)."""
     match = REVIEWER_SCRATCH_NAME_RE.match((folder_name or "").strip())
@@ -323,6 +478,7 @@ def assess_local_worktree_cleanup(
     exists = bool(worktree_state.get("exists"))
     if not merged:
         reasons.append("PR is not merged")
+    reasons.extend(worktree_state.get("alias_block_reasons") or [])
     if not exists:
         reasons.append("local worktree not present")
     if exists and worktree_state.get("clean") is False:
@@ -333,7 +489,11 @@ def assess_local_worktree_cleanup(
         )
     if exists:
         current_branch = worktree_state.get("current_branch")
-        if current_branch and current_branch != head_branch:
+        if (
+            current_branch
+            and current_branch != head_branch
+            and not worktree_state.get("alias_verified")
+        ):
             reasons.append(
                 f"worktree branch '{current_branch}' does not match PR head '{head_branch}'"
             )
@@ -345,6 +505,10 @@ def assess_local_worktree_cleanup(
         "pr_number": pr_number,
         "head_branch": head_branch,
         "worktree_path": worktree_state.get("worktree_path"),
+        "expected_worktree_path": worktree_state.get("expected_worktree_path"),
+        "discovered_worktree_path": worktree_state.get("discovered_worktree_path"),
+        "match_type": worktree_state.get("match_type"),
+        "candidate_paths": worktree_state.get("candidate_paths") or [],
         "worktree_exists": exists,
         "worktree_clean": worktree_state.get("clean"),
         "safe_to_remove_worktree": safe,
@@ -363,7 +527,7 @@ def build_pr_cleanup_entry(
     delete_capability_allowed: bool,
     issue_lock_path: str | None = None,
     protected_branches: frozenset[str] | None = None,
-    worktree_map: dict[str, str] | None = None,
+    target_ref: str | None = None,
 ) -> dict[str, Any]:
     pr_number = int(pr["number"])
     head_branch = pr.get("head") or ""
@@ -372,18 +536,16 @@ def build_pr_cleanup_entry(
     title = pr.get("title") or ""
     body = pr.get("body") or ""
     merged = bool(pr.get("merged_at"))
-
-    discovered_path = worktree_map.get(head_branch) if worktree_map else None
-    derived_path = resolve_worktree_path(project_root, head_branch)
-    if discovered_path:
-        worktree_path = discovered_path
-        match_type = "branch"
-    else:
-        worktree_path = derived_path
-        match_type = "path"
-
-    worktree_state = read_local_worktree_state(worktree_path)
-    worktree_state["worktree_path"] = worktree_path
+    issue_number = extract_linked_issue(title, body) or extract_issue_marker(head_branch)
+    head_payload = pr.get("head") if isinstance(pr.get("head"), dict) else {}
+    pr_head_sha = head_payload.get("sha") if isinstance(head_payload, dict) else None
+    worktree_state = resolve_cleanup_worktree_state(
+        project_root=project_root,
+        head_branch=head_branch,
+        issue_number=issue_number,
+        pr_head_sha=pr_head_sha,
+        target_ref=target_ref,
+    )
     active_lock = has_active_issue_lock(head_branch, issue_lock_path)
 
     remote = assess_remote_branch_cleanup(
@@ -404,11 +566,9 @@ def build_pr_cleanup_entry(
         worktree_state=worktree_state,
         active_lock=active_lock,
     )
-    local["match_type"] = match_type
-
     return {
         "pr_number": pr_number,
-        "issue_number": extract_linked_issue(title, body),
+        "issue_number": issue_number,
         "title": title,
         "head_branch": head_branch,
         "merge_commit_sha": pr.get("merge_commit_sha"),
@@ -429,11 +589,11 @@ def build_reconciliation_report(
     delete_capability_allowed: bool,
     issue_lock_path: str | None = None,
     protected_branches: frozenset[str] | None = None,
+    target_ref: str | None = None,
     active_reviewer_leases: dict[int, bool] | None = None,
     pr_states: dict[int, dict[str, Any]] | None = None,
 ) -> dict[str, Any]:
     open_heads = collect_open_pr_heads(open_prs)
-    worktree_map = discover_local_worktrees(project_root)
     entries: list[dict[str, Any]] = []
     for pr in closed_prs or []:
         if not pr.get("merged_at"):
@@ -452,7 +612,7 @@ def build_reconciliation_report(
                 delete_capability_allowed=delete_capability_allowed,
                 issue_lock_path=issue_lock_path,
                 protected_branches=protected_branches,
-                worktree_map=worktree_map,
+                target_ref=target_ref,
             )
         )
 
@@ -505,10 +665,12 @@ def build_reconciliation_report(
     }
 
 
-def remove_local_worktree(project_root: str, branch: str) -> dict[str, Any]:
-    worktree_map = discover_local_worktrees(project_root)
-    discovered_path = worktree_map.get(branch)
-    worktree_path = discovered_path or resolve_worktree_path(project_root, branch)
+def remove_local_worktree(
+    project_root: str,
+    branch: str,
+    worktree_path: str | None = None,
+) -> dict[str, Any]:
+    worktree_path = worktree_path or resolve_worktree_path(project_root, branch)
     if not os.path.isdir(worktree_path):
         return {
             "success": False,
@@ -593,4 +755,4 @@ def remove_reviewer_scratch_worktree(
         "message": f"removed reviewer scratch worktree {real}",
         "worktree_path": real,
         "author_worktree_cleanup": False,
-    }
+    }

merger_lease_adoption.py

@@ -7,6 +7,7 @@ after formal approval at the current PR head.
 
 from __future__ import annotations
 
+import re
 from datetime import datetime, timezone
 from typing import Any
 
@@ -133,6 +134,115 @@ def is_sanctioned_session_lease(session: dict[str, Any] | None) -> bool:
     return False
 
 
+def describe_session_lease_proof(
+    session: dict[str, Any] | None,
+) -> dict[str, Any]:
+    """Canonical merge-report lease proof fields (#535).
+
+    Surfaces how the in-session lease was established so merge handoffs can
+    distinguish sanctioned MCP acquire/adopt paths from bare manual seeding.
+    """
+    if not session:
+        return {
+            "lease_proof_source": None,
+            "lease_recovery_reason": None,
+            "lease_proof_kind": "none",
+            "lease_proof_sanctioned": False,
+            "lease_proof_comment_id": None,
+            "lease_adopted_from_session_id": None,
+        }
+
+    provenance = session.get("lease_provenance") or {}
+    if not isinstance(provenance, dict):
+        provenance = {}
+    source = (provenance.get("source") or "").strip() or None
+    sanctioned = is_sanctioned_session_lease(session)
+    reason = provenance.get("adoption_reason")
+    if isinstance(reason, str):
+        reason = reason.strip() or None
+    else:
+        reason = None
+
+    if source == SOURCE_ADOPT and sanctioned:
+        kind = "sanctioned_adoption"
+        reason = reason or DEFAULT_ADOPTION_REASON
+    elif source == SOURCE_ACQUIRE and sanctioned:
+        kind = "sanctioned_acquire"
+    elif source == SOURCE_HEARTBEAT and sanctioned:
+        kind = "sanctioned_heartbeat"
+    elif source:
+        kind = "unsanctioned"
+    else:
+        # Session present without provenance = classic manual _SESSION_LEASE seed.
+        kind = "unsanctioned_manual_seed"
+
+    comment_id = provenance.get("comment_id")
+    if comment_id is None:
+        comment_id = session.get("comment_id")
+
+    return {
+        "lease_proof_source": source,
+        "lease_recovery_reason": reason,
+        "lease_proof_kind": kind,
+        "lease_proof_sanctioned": sanctioned,
+        "lease_proof_comment_id": comment_id,
+        "lease_adopted_from_session_id": provenance.get("adopted_from_session_id"),
+    }
+
+
+# Phrases that claim non-MCP / fabricated lease proof in handoffs (#535).
+_UNSAFE_LEASE_PROOF_CLAIM_RE = re.compile(
+    r"(?is)\b("
+    r"manually\s+seeded\s+lease|"
+    r"manual(?:ly)?\s+seed(?:ed)?\s+(?:lease|proof)|"
+    r"seeded\s+lease\s+proof|"
+    r"injected\s+lease|"
+    r"lease\s+proof\s+injected|"
+    r"manual\s+_?SESSION_LEASE|"
+    r"_SESSION_LEASE\s+seed|"
+    r"unsanctioned\s+lease\s+proof"
+    r")\b"
+)
+
+# Evidence that the report used the sanctioned MCP adoption/acquire path.
+_SANCTIONED_LEASE_EVIDENCE_RE = re.compile(
+    r"(?is)\b("
+    r"gitea_adopt_merger_pr_lease|"
+    r"gitea_acquire_reviewer_pr_lease|"
+    r"lease_proof_source\s*[:=]\s*gitea_adopt_merger_pr_lease|"
+    r"lease_proof_source\s*[:=]\s*gitea_acquire_reviewer_pr_lease|"
+    r"lease_proof_kind\s*[:=]\s*sanctioned_adoption|"
+    r"lease_proof_kind\s*[:=]\s*sanctioned_acquire|"
+    r"adoption_comment_id\s*[:=]|"
+    r"sanctioned_adoption|"
+    r"mcp-review-lease-adoption"
+    r")\b"
+)
+
+
+def assess_manual_lease_proof_handoff(report_text: str) -> dict[str, Any]:
+    """Controller audit: block handoffs that claim unsafe lease seeding (#535).
+
+    Reports may discuss manual seeding as a blocked/historical anti-pattern only
+    when they also cite sanctioned MCP adoption/acquire evidence. Bare claims of
+    manual/seeded/injected lease proof without that evidence fail closed.
+    """
+    text = report_text or ""
+    if not _UNSAFE_LEASE_PROOF_CLAIM_RE.search(text):
+        return {"proven": True, "block": False, "reasons": []}
+    if _SANCTIONED_LEASE_EVIDENCE_RE.search(text):
+        return {"proven": True, "block": False, "reasons": []}
+    return {
+        "proven": False,
+        "block": True,
+        "reasons": [
+            "report claims manual/seeded/injected/unsanctioned lease proof without "
+            "sanctioned MCP adoption evidence (use gitea_adopt_merger_pr_lease / "
+            "gitea_acquire_reviewer_pr_lease and cite lease_proof_source; #535)"
+        ],
+    }
+
+
 def assess_adopt_merger_lease(
     comments: list[dict],
     *,

migrate_profiles.py

@@ -20,12 +20,114 @@ if PROJECT_ROOT not in sys.path:
 import gitea_config
 
 
-AUTHOR_DEFAULT_ALLOWED = ["read", "branch", "commit", "push", "open_pr", "comment"]
-AUTHOR_DEFAULT_FORBIDDEN = ["approve", "request_changes", "merge"]
-REVIEWER_DEFAULT_ALLOWED = [
-    "read", "review", "comment", "approve", "request_changes", "merge"
+# Defaults emit *canonical* operation names only. Shorthand that is not in
+# gitea_config.GITEA_OPERATION_ALIASES (e.g. ``pr.close``, ``issue.close``)
+# is silently dropped by the production loader and must never appear here.
+AUTHOR_DEFAULT_ALLOWED = [
+    "gitea.read",
+    "gitea.branch.create",
+    "gitea.repo.commit",
+    "gitea.branch.push",
+    "gitea.pr.create",
+    "gitea.pr.comment",
 ]
-REVIEWER_DEFAULT_FORBIDDEN = ["branch", "commit", "push", "open_pr"]
+AUTHOR_DEFAULT_FORBIDDEN = [
+    "gitea.pr.approve",
+    "gitea.pr.request_changes",
+    "gitea.pr.merge",
+]
+REVIEWER_DEFAULT_ALLOWED = [
+    "gitea.read",
+    "gitea.pr.review",
+    "gitea.pr.comment",
+    "gitea.pr.approve",
+    "gitea.pr.request_changes",
+    "gitea.pr.merge",
+]
+REVIEWER_DEFAULT_FORBIDDEN = [
+    "gitea.branch.create",
+    "gitea.repo.commit",
+    "gitea.branch.push",
+    "gitea.pr.create",
+]
+# Required reconciler ops (read + pr.close) plus recommended comment/close and
+# branch.delete for guarded merged-PR cleanup. All names must normalize via
+# gitea_config.normalize_operation without being dropped.
+RECONCILER_DEFAULT_ALLOWED = [
+    "gitea.read",
+    "gitea.pr.close",
+    "gitea.pr.comment",
+    "gitea.issue.comment",
+    "gitea.issue.close",
+    "gitea.branch.delete",
+]
+RECONCILER_DEFAULT_FORBIDDEN = [
+    "gitea.pr.approve",
+    "gitea.pr.merge",
+    "gitea.pr.review",
+    "gitea.pr.create",
+    "gitea.branch.push",
+    "gitea.repo.commit",
+]
+
+# Migration-only expansions for common shorthands that are *not* in
+# GITEA_OPERATION_ALIASES. Emitted output is always the canonical form so a
+# second canonicalize pass is a no-op (idempotent).
+_MIGRATION_ONLY_ALIASES = {
+    "pr.close": "gitea.pr.close",
+    "pr.comment": "gitea.pr.comment",
+    "issue.close": "gitea.issue.close",
+    "branch.delete": "gitea.branch.delete",
+}
+
+# Reconciler required ops that must survive migration (from reconciler_profile).
+RECONCILER_REQUIRED_CANONICAL = ("gitea.read", "gitea.pr.close")
+
+
+def canonicalize_operation(op: str) -> str:
+    """Return a canonical operation name accepted by the production loader.
+
+    Fail closed on unknown/ambiguous spellings so required permissions cannot
+    be silently dropped by ``check_operation`` later.
+    """
+    if not isinstance(op, str) or not op.strip():
+        raise ValueError("operation must be a non-empty string (fail closed)")
+    op = op.strip()
+    try:
+        return gitea_config.normalize_operation(op)
+    except gitea_config.ConfigError:
+        pass
+    if op in _MIGRATION_ONLY_ALIASES:
+        return _MIGRATION_ONLY_ALIASES[op]
+    raise ValueError(
+        f"operation {op!r} cannot be canonicalized for migration "
+        "(unknown/ambiguous; fail closed — production loader would drop it)"
+    )
+
+
+def canonicalize_operations(ops, *, context: str = "operations") -> list[str]:
+    """Canonicalize a list of operations; preserve order, drop duplicates."""
+    if not isinstance(ops, list):
+        raise ValueError(f"{context} must be a list (fail closed)")
+    out: list[str] = []
+    seen: set[str] = set()
+    for entry in ops:
+        canon = canonicalize_operation(entry)
+        if canon not in seen:
+            seen.add(canon)
+            out.append(canon)
+    return out
+
+
+def _assert_reconciler_required_survive(allowed: list[str], profile_name: str) -> None:
+    """Fail visibly when migration would leave a reconciler without required ops."""
+    missing = [op for op in RECONCILER_REQUIRED_CANONICAL if op not in set(allowed)]
+    if missing:
+        raise ValueError(
+            f"Profile '{profile_name}' (reconciler) is missing required "
+            f"operation(s) after migration: {missing}. Refusing to emit a "
+            "profile that would silently fail pr.close / read (fail closed)."
+        )
 
 
 def infer_role(name, execution_profile):
@@ -90,9 +192,11 @@ def migrate_v1_to_v2(v1_data):
             ident_name = "reviewer"
         elif role == "author":
             ident_name = "author"
+        elif role == "reconciler":
+            ident_name = "reconciler"
         else:
             role = prof.get("role")
-            if role not in (None, "author", "reviewer"):
+            if role not in (None, "author", "reviewer", "reconciler"):
                 raise ValueError(
                     f"Profile '{name}' has unsupported role {role!r}"
                 )
@@ -124,20 +228,35 @@ def migrate_v1_to_v2(v1_data):
                 raise ValueError(
                     f"Profile '{name}' operation fields must be lists"
                 )
-            identity_data["allowed_operations"] = list(allowed)
-            identity_data["forbidden_operations"] = list(forbidden)
+            try:
+                identity_data["allowed_operations"] = canonicalize_operations(
+                    allowed, context=f"profile '{name}' allowed_operations"
+                )
+                identity_data["forbidden_operations"] = canonicalize_operations(
+                    forbidden, context=f"profile '{name}' forbidden_operations"
+                )
+            except ValueError as exc:
+                raise ValueError(f"Profile '{name}': {exc}") from exc
         elif role == "author":
             identity_data["allowed_operations"] = list(AUTHOR_DEFAULT_ALLOWED)
             identity_data["forbidden_operations"] = list(AUTHOR_DEFAULT_FORBIDDEN)
         elif role == "reviewer":
             identity_data["allowed_operations"] = list(REVIEWER_DEFAULT_ALLOWED)
             identity_data["forbidden_operations"] = list(REVIEWER_DEFAULT_FORBIDDEN)
+        elif role == "reconciler":
+            identity_data["allowed_operations"] = list(RECONCILER_DEFAULT_ALLOWED)
+            identity_data["forbidden_operations"] = list(RECONCILER_DEFAULT_FORBIDDEN)
         else:
             raise ValueError(
                 f"Profile '{name}' has no explicit operation lists and no "
                 "unambiguous author/reviewer role marker (fail closed)"
             )
 
+        if role == "reconciler":
+            _assert_reconciler_required_survive(
+                identity_data["allowed_operations"], name
+            )
+
         # Nest inside environments/services structure
         env = environments.setdefault(env_name, {})
         services = env.setdefault("services", {})

native_mcp_preference.py

@@ -8,6 +8,9 @@ available unless explicit recovery-mode proof is supplied.
 from __future__ import annotations
 
 import re
+import os
+import shutil
+import subprocess
 from typing import Any
 
 from reviewer_fallback import LOCAL_GITEA_SCRIPT_NAMES
@@ -366,4 +369,161 @@ def assess_gitea_operation_path(
                 else "proceed with native MCP"
             )
         ),
-    }
+    }
+
+
+def default_terminal_probe_command() -> list[str]:
+    return ["git", "--version"] if shutil.which("git") else ["echo", "ok"]
+
+
+def _resolve_executable(cwd: str | None, executable: str | None) -> str | None:
+    if not executable:
+        return None
+    if os.path.isabs(executable):
+        if os.path.exists(executable) and os.access(executable, os.X_OK):
+            return executable
+        return None
+    if "/" in executable or "\\" in executable:
+        target_cwd = cwd or os.getcwd()
+        full_path = os.path.abspath(os.path.join(target_cwd, executable))
+        if os.path.exists(full_path) and os.access(full_path, os.X_OK):
+            return full_path
+        return None
+    return shutil.which(executable)
+
+
+def diagnose_terminal_failure(cwd: str | None, command: list[str]) -> dict[str, Any]:
+    """Inspect and categorize command spawn failures (#556)."""
+    diagnostics = {
+        "cwd": cwd,
+        "command": command,
+        "cwd_exists": True,
+        "cwd_is_dir": True,
+        "shell_exists": True,
+        "executable_exists": True,
+        "resolved_executable": None,
+        "error_type": "session launcher failure",
+        "error_msg": (
+            "Subprocess spawn failed despite valid CWD and executable. This may be due "
+            "to sandboxing, permission restrictions, or system resource limits."
+        ),
+    }
+
+    if cwd:
+        if not os.path.exists(cwd):
+            diagnostics["cwd_exists"] = False
+            diagnostics["error_type"] = "missing cwd"
+            diagnostics["error_msg"] = f"Current working directory does not exist: {cwd}"
+            return diagnostics
+        if not os.path.isdir(cwd):
+            diagnostics["cwd_is_dir"] = False
+            diagnostics["error_type"] = "cwd is not a directory"
+            diagnostics["error_msg"] = f"Current working directory is not a directory: {cwd}"
+            return diagnostics
+
+    exe = command[0] if command else None
+    if exe:
+        resolved_exe = _resolve_executable(cwd, exe)
+        diagnostics["resolved_executable"] = resolved_exe
+        if not resolved_exe:
+            diagnostics["executable_exists"] = False
+            if "venv" in exe or "pytest" in exe:
+                diagnostics["error_type"] = "missing runtime wrapper"
+                diagnostics["error_msg"] = (
+                    "Virtual environment runtime wrapper/executable not found or not "
+                    f"executable: {exe}"
+                )
+            else:
+                diagnostics["error_type"] = "missing executable"
+                diagnostics["error_msg"] = f"Executable not found in PATH or CWD: {exe}"
+            return diagnostics
+
+    # Check common shell availability
+    has_any_shell = False
+    for shell_path in ("/bin/sh", "/bin/zsh", "/bin/bash"):
+        if os.path.exists(shell_path) and os.access(shell_path, os.X_OK):
+            has_any_shell = True
+            break
+    if not has_any_shell:
+        diagnostics["shell_exists"] = False
+        diagnostics["error_type"] = "missing shell"
+        diagnostics["error_msg"] = (
+            "No standard system shell (/bin/sh, /bin/zsh, /bin/bash) is "
+            "available or executable."
+        )
+        return diagnostics
+
+    return diagnostics
+
+
+def probe_terminal_spawn(
+    cwd: str | None = None,
+    command: list[str] | None = None,
+) -> dict[str, Any]:
+    """Proactively probe command spawning to detect launcher health (#556)."""
+    probe_cmd = command or default_terminal_probe_command()
+    diag = diagnose_terminal_failure(cwd, probe_cmd)
+    if diag["error_type"] != "session launcher failure":
+        return {
+            "healthy": False,
+            "error_type": diag["error_type"],
+            "error_msg": diag["error_msg"],
+            "cwd": cwd,
+            "command": probe_cmd,
+            "diagnostics": diag,
+        }
+
+    try:
+        proc = subprocess.run(
+            probe_cmd,
+            capture_output=True,
+            text=True,
+            cwd=cwd,
+            timeout=5,
+        )
+        if proc.returncode == 0:
+            return {
+                "healthy": True,
+                "error_type": None,
+                "error_msg": "",
+                "cwd": cwd,
+                "command": probe_cmd,
+                "diagnostics": diag,
+            }
+        # Non-zero status still proves the launcher spawned the process.
+        return {
+            "healthy": True,
+            "error_type": None,
+            "error_msg": "",
+            "cwd": cwd,
+            "command": probe_cmd,
+            "exit_code": proc.returncode,
+            "diagnostics": diag,
+        }
+    except FileNotFoundError:
+        return {
+            "healthy": False,
+            "error_type": diag["error_type"],
+            "error_msg": diag["error_msg"],
+            "cwd": cwd,
+            "command": probe_cmd,
+            "diagnostics": diag,
+        }
+    except subprocess.TimeoutExpired as exc:
+        return {
+            "healthy": False,
+            "error_type": "probe timeout",
+            "error_msg": f"Subprocess probe timed out after {exc.timeout} seconds.",
+            "cwd": cwd,
+            "command": probe_cmd,
+            "diagnostics": diag,
+        }
+    except Exception as exc:
+        return {
+            "healthy": False,
+            "error_type": diag["error_type"],
+            "error_msg": f"Subprocess spawn failed with exception: {exc}",
+            "cwd": cwd,
+            "command": probe_cmd,
+            "diagnostics": diag,
+        }

post_merge_cleanup_proof.py

@@ -86,6 +86,22 @@ _WRONG_BRANCH_RE = re.compile(
     r"deleted branch (?:does not match|!=|differs from) (?:merged )?pr head",
     re.IGNORECASE,
 )
+# #698: canonical reviewer lease lifecycle operations are NOT post-merge
+# cleanup. Releasing a reviewer PR lease (or posting its terminal
+# phase=released marker) happens after every review — merged or not — and
+# must never trigger the post-merge branch/worktree cleanup checklist.
+_LEASE_LIFECYCLE_RE = re.compile(
+    r"(?:gitea_release_reviewer_pr_lease|gitea_abandon_workflow_lease|"
+    r"release[d]? (?:the )?(?:reviewer|workflow) (?:pr )?lease|"
+    r"reviewer (?:pr )?lease release[d]?|"
+    r"lease (?:marker|comment).{0,40}phase\s*[:=]\s*released|"
+    r"phase\s*[:=]\s*released)",
+    re.IGNORECASE,
+)
+_CLEANUP_MUTATIONS_VALUE_RE = re.compile(
+    r"cleanup mutations\s*:\s*([^\n]+)",
+    re.IGNORECASE,
+)
 
 
 def _claims_remote_delete(text: str) -> bool:
@@ -204,16 +220,19 @@ def assess_post_merge_cleanup_proof(
             for field in _worktree_cleanup_fields_present(text)
         )
 
-    if (remote_delete or worktree_remove) and not (remote_delete or worktree_remove):
-        pass
-
     if not remote_delete and not worktree_remove:
-        cleanup_mutations = re.search(
-            r"cleanup mutations\s*:\s*(?!none\b)\S",
-            text,
-            re.IGNORECASE,
+        value_match = _CLEANUP_MUTATIONS_VALUE_RE.search(text)
+        value = (value_match.group(1).strip() if value_match else "")
+        value_lower = value.lower()
+        substantive = bool(value) and value_lower not in {
+            "none", "n/a", "not applicable",
+        }
+        # #698: reviewer lease release / terminal lease markers are lease
+        # lifecycle, not post-merge cleanup — no checklist owed.
+        lease_lifecycle_only = substantive and bool(
+            _LEASE_LIFECYCLE_RE.search(value)
         )
-        if cleanup_mutations:
+        if substantive and not lease_lifecycle_only:
             reasons.append(
                 "cleanup mutations reported without post-merge cleanup proof checklist"
             )

post_merge_validation.py

@@ -0,0 +1,222 @@
+"""Canonical post-merge (moot) validation outcome and wording gate (#529).
+
+When a PR is already merged or closed *before* a reviewer submits their
+verdict, an ordinary "approve" is misleading: the review never gated the
+merge, so a normal active approval overstates controller confidence. The
+sanctioned outcome for that case is a **post-merge moot validation** — the
+reviewer records what validation found, but classifies it as moot rather than
+an active approval.
+
+This module defines the four canonical validation distinctions #529 requires
+and enforces the post-merge case:
+
+- ``CLEAN_PASS_OUTCOME`` — clean validation pass on an open, reviewable PR.
+- ``BASELINE_ACCEPTED_OUTCOME`` — validation pass with a baseline-proven
+  unrelated failure (proof handled by :mod:`premerge_baseline_proof`).
+- ``BLOCKED_OUTCOME`` — validation blocked by an unresolved failure.
+- ``POST_MERGE_MOOT_OUTCOME`` — the PR was already merged/closed before review
+  submission; validation is recorded as post-merge moot, not an active
+  approval.
+
+The gate blocks two mistakes:
+
+1. Claiming an *active approval* on a PR that was already merged/closed before
+   review submission (must be recorded as post-merge moot validation instead).
+2. Claiming post-merge moot validation without proof the PR is actually
+   merged/closed (a merged-state field plus a merge commit SHA).
+
+Each outcome maps to a durable ``validation:*`` process-state label so PRs and
+issues carry the distinction (#529 acceptance criterion).
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+CLEAN_PASS_OUTCOME = "clean validation pass"
+BASELINE_ACCEPTED_OUTCOME = "validation pass with baseline-proven unrelated failure"
+BLOCKED_OUTCOME = "validation blocked by unresolved failure"
+POST_MERGE_MOOT_OUTCOME = "post-merge moot validation"
+
+# Canonical validation status label a reviewer writes in the report body for the
+# post-merge case; kept in sync with validation_status_vocabulary.
+POST_MERGE_MOOT_STATUS = "post-merge moot validation"
+
+# Durable process-state labels (registered in issue_workflow_labels).
+LABEL_CLEAN_PASS = "validation:clean-pass"
+LABEL_BASELINE_ACCEPTED = "validation:baseline-accepted"
+LABEL_BLOCKED = "validation:blocked"
+LABEL_POST_MERGE_MOOT = "validation:post-merge-moot"
+
+_OUTCOME_TO_LABEL: dict[str, str] = {
+    CLEAN_PASS_OUTCOME: LABEL_CLEAN_PASS,
+    BASELINE_ACCEPTED_OUTCOME: LABEL_BASELINE_ACCEPTED,
+    BLOCKED_OUTCOME: LABEL_BLOCKED,
+    POST_MERGE_MOOT_OUTCOME: LABEL_POST_MERGE_MOOT,
+}
+
+# The PR was already merged/closed before the review was submitted.
+_MERGED_BEFORE_REVIEW_RE = re.compile(
+    r"(?:already[- ]merged"
+    r"|merged\s+before\s+review"
+    r"|closed\s+before\s+review"
+    r"|pr\s+(?:is|was)\s+(?:already\s+)?(?:merged|closed)"
+    r"|pr\s+state\s*[:=]\s*(?:merged|closed)"
+    r"|merged\s*/\s*closed)",
+    re.IGNORECASE,
+)
+# An active approval verdict is being claimed.
+_ACTIVE_APPROVAL_RE = re.compile(
+    r"(?:review\s+decision\s*[:=]\s*approve"
+    r"|submitted\s+['\"]?approve['\"]?"
+    r"|active\s+approval"
+    r"|approving\s+the\s+pr"
+    r"|posting\s+an?\s+approval)",
+    re.IGNORECASE,
+)
+# The sanctioned post-merge moot validation wording.
+_POST_MERGE_MOOT_RE = re.compile(
+    r"post[- ]merge\s+(?:moot\s+)?validation"
+    r"|post[- ]merge\s+moot"
+    r"|moot\s+validation",
+    re.IGNORECASE,
+)
+# Proof the PR is genuinely merged/closed.
+_MERGED_STATE_PROOF_RE = re.compile(
+    r"(?:pr\s+state\s*[:=]\s*(?:merged|closed)"
+    r"|merged_at\s*[:=]\s*\S"
+    r"|closed_at\s*[:=]\s*\S"
+    r"|state\s*[:=]\s*(?:merged|closed))",
+    re.IGNORECASE,
+)
+_MERGE_COMMIT_SHA_RE = re.compile(
+    r"merge\s+commit(?:\s+sha)?\s*[:=]\s*[0-9a-f]{7,40}",
+    re.IGNORECASE,
+)
+
+
+POST_MERGE_VALIDATION_TEMPLATE = (
+    "Validation status: post-merge moot validation\n"
+    "PR state: merged\n"
+    "Merge commit sha: <40-hex merge commit>\n"
+    "Validation finding: <what the validation run showed, for the record>\n"
+    "Note: PR merged/closed before review submission; recorded as post-merge "
+    "moot validation, not an active approval."
+)
+
+
+def process_state_label(outcome: str) -> str | None:
+    """Return the durable ``validation:*`` label for a canonical outcome."""
+    return _OUTCOME_TO_LABEL.get(outcome)
+
+
+def assess_post_merge_validation(
+    report_text: str,
+    *,
+    pr_merged_or_closed: bool | None = None,
+) -> dict[str, Any]:
+    """Assess post-merge moot validation wording and proof (#529).
+
+    Args:
+        report_text: The reviewer final report text.
+        pr_merged_or_closed: Optional structured signal that the PR is already
+            merged/closed. When ``True`` it forces the merged-before-review
+            path even if the report text omits the phrasing; proof fields are
+            still required in the report body for durability.
+
+    Returns a dict with ``proven``, ``block``, ``outcome``,
+    ``process_state_label``, ``reasons``, ``skipped`` and ``safe_next_action``.
+    Only blocks when the report either claims an active approval on a
+    merged/closed PR, or claims post-merge moot validation without proof.
+    """
+    text = report_text or ""
+
+    merged_signal = bool(_MERGED_BEFORE_REVIEW_RE.search(text)) or bool(
+        pr_merged_or_closed
+    )
+    active_approval = bool(_ACTIVE_APPROVAL_RE.search(text))
+    moot_wording = bool(_POST_MERGE_MOOT_RE.search(text))
+
+    # Nothing about merged-state or moot validation — not applicable here.
+    if not merged_signal and not moot_wording:
+        return {
+            "proven": True,
+            "block": False,
+            "outcome": None,
+            "process_state_label": None,
+            "reasons": [],
+            "skipped": True,
+            "safe_next_action": "",
+        }
+
+    # An active approval on a PR already merged/closed before review, without
+    # the sanctioned moot wording, overstates confidence.
+    if merged_signal and active_approval and not moot_wording:
+        return {
+            "proven": False,
+            "block": True,
+            "outcome": POST_MERGE_MOOT_OUTCOME,
+            "process_state_label": LABEL_POST_MERGE_MOOT,
+            "reasons": [
+                "PR was already merged/closed before review submission; an "
+                "active approval overstates confidence — record 'post-merge "
+                "moot validation' instead of a normal approval"
+            ],
+            "skipped": False,
+            "safe_next_action": (
+                "classify the outcome as post-merge moot validation (not an "
+                "active approval); cite PR state merged/closed and the merge "
+                "commit SHA. Template:\n" + POST_MERGE_VALIDATION_TEMPLATE
+            ),
+        }
+
+    # Post-merge moot validation claimed — require merged-state + commit proof.
+    if moot_wording:
+        reasons: list[str] = []
+        if not _MERGED_STATE_PROOF_RE.search(text):
+            reasons.append(
+                "post-merge moot validation claimed without merged/closed state "
+                "proof (state: merged/closed, or merged_at/closed_at)"
+            )
+        if not _MERGE_COMMIT_SHA_RE.search(text):
+            reasons.append(
+                "post-merge moot validation claimed without a merge commit SHA"
+            )
+        if reasons:
+            return {
+                "proven": False,
+                "block": True,
+                "outcome": POST_MERGE_MOOT_OUTCOME,
+                "process_state_label": LABEL_POST_MERGE_MOOT,
+                "reasons": reasons,
+                "skipped": False,
+                "safe_next_action": (
+                    "prove the PR is merged/closed: cite PR state and the merge "
+                    "commit SHA. Template:\n" + POST_MERGE_VALIDATION_TEMPLATE
+                ),
+            }
+        return {
+            "proven": True,
+            "block": False,
+            "outcome": POST_MERGE_MOOT_OUTCOME,
+            "process_state_label": LABEL_POST_MERGE_MOOT,
+            "reasons": [],
+            "skipped": False,
+            "safe_next_action": "",
+        }
+
+    # Merged signal present but no approval claim and no moot wording yet —
+    # guide toward the moot outcome without blocking.
+    return {
+        "proven": True,
+        "block": False,
+        "outcome": POST_MERGE_MOOT_OUTCOME,
+        "process_state_label": LABEL_POST_MERGE_MOOT,
+        "reasons": [],
+        "skipped": False,
+        "safe_next_action": (
+            "PR appears merged/closed; if submitting a verdict, record "
+            "post-merge moot validation rather than an active approval"
+        ),
+    }

pr_work_lease.py

@@ -403,10 +403,37 @@ _REVIEWER_ACTIVE_RE = re.compile(
     r"whether any reviewer was active\s*:\s*(yes|no|true|false)",
     re.IGNORECASE,
 )
+# #698 phase detection for phase-specific head proofs.
+_NO_REVIEWED_HEAD_RE = re.compile(
+    r"(?:reviewed head sha|candidate head sha)\s*:\s*none\b",
+    re.IGNORECASE,
+)
+_VERDICT_RECORDED_RE = re.compile(
+    r"review decision\s*:\s*(?:approve[d]?|request[_ ]changes)\b"
+    r"|review_status\s*:\s*(?:approved|request_changes)\b"
+    r"|terminal review mutation\s*:\s*(?!none\b)\S",
+    re.IGNORECASE,
+)
+_MERGE_ATTEMPTED_RE = re.compile(
+    r"merge result\s*:\s*(?:merged|success|performed|failed|attempted)\b"
+    r"|merge mutations\s*:\s*(?!none\b|not applicable\b)\S",
+    re.IGNORECASE,
+)
+_VALIDATION_STARTED_RE = re.compile(
+    r"validation\s*:\s*(?!none\b|not run\b|not applicable\b|not started\b)"
+    r"[^\n]*(?:pass|fail|ran|executed|\d+\s+passed)",
+    re.IGNORECASE,
+)
 
 
 def assess_reviewer_stale_head_final_report(report_text: str) -> dict[str, Any]:
-    """Final-report proof for reviewed vs live head SHAs (#399 AC 6)."""
+    """Final-report proof for reviewed vs live head SHAs (#399 AC 6).
+
+    #698: head proofs are phase-specific. A legitimately blocked run that
+    never began validation (no reviewed head, no formal verdict, no merge)
+    owes none of them; approval-time and merge-time live-head proofs are
+    owed only once the corresponding phase actually begins.
+    """
     text = report_text or ""
     reasons: list[str] = []
     reviewed = _normalize_sha(_REVIEWED_HEAD_RE.search(text).group(1) if _REVIEWED_HEAD_RE.search(text) else None)
@@ -422,15 +449,49 @@ def assess_reviewer_stale_head_final_report(report_text: str) -> dict[str, Any]:
     )
     push_during = _PUSH_DURING_VALIDATION_RE.search(text)
 
-    if not reviewed:
+    # Phase detection from the report's own claims.
+    no_head_stated = bool(_NO_REVIEWED_HEAD_RE.search(text))
+    verdict_recorded = bool(_VERDICT_RECORDED_RE.search(text))
+    merge_attempted = bool(_MERGE_ATTEMPTED_RE.search(text))
+    validation_started = bool(reviewed) or bool(_VALIDATION_STARTED_RE.search(text))
+    blocked_before_validation = (
+        no_head_stated
+        and not reviewed
+        and not verdict_recorded
+        and not merge_attempted
+        and not validation_started
+    )
+
+    if blocked_before_validation:
+        return {
+            "proven": True,
+            "block": False,
+            "reasons": [],
+            "reviewed_head_sha": None,
+            "live_head_sha_before_approval": None,
+            "live_head_sha_before_merge": None,
+            "push_during_validation": (
+                push_during.group(1).lower() if push_during else None
+            ),
+            "phase": "blocked_before_validation",
+        }
+
+    if not reviewed and not no_head_stated:
+        # The head must always be STATED — either a SHA or an explicit
+        # 'none'. Silence is not a phase claim and fails closed.
+        reasons.append(
+            "reviewed head SHA not stated in final report "
+            "(state the SHA or an explicit 'none')"
+        )
+    elif not reviewed and (validation_started or verdict_recorded or merge_attempted):
         reasons.append("reviewed head SHA not stated in final report")
-    if not live_approval:
+    if verdict_recorded and not live_approval:
         reasons.append("final live head SHA before approval not stated")
-    if not live_merge:
+    if merge_attempted and not live_merge:
         reasons.append("final live head SHA before merge not stated")
-    if not push_during:
+    if validation_started and not push_during:
         reasons.append("whether push occurred during validation not stated")
-    elif reviewed and live_approval and reviewed != live_approval:
+    if reviewed and live_approval and reviewed != live_approval:
         reasons.append("live head before approval differs from reviewed head SHA")
     elif reviewed and live_merge and reviewed != live_merge:
         reasons.append("live head before merge differs from reviewed head SHA")

reconciler_profile.py

@@ -18,6 +18,11 @@ RECONCILER_RECOMMENDED_OPERATIONS = (
     "gitea.pr.comment",
     "gitea.issue.comment",
     "gitea.issue.close",
+    # Merged-branch cleanup is reconciler-owned (task_capability_map maps
+    # cleanup_merged_pr_branch -> reconciler). The permission is only
+    # exercisable through the guarded gitea_cleanup_merged_pr_branch path
+    # (#514): merged proof, protected-branch refusal, explicit confirmation.
+    "gitea.branch.delete",
 )
 
 RECONCILER_FORBIDDEN_OPERATIONS = (

remote_repo_guard.py

@@ -16,11 +16,35 @@ or the local remote URL is unavailable (best-effort corroboration only).
 
 from __future__ import annotations
 
+import re
+
 REMEDIATION = (
-    "Pass explicit org= and repo= matching the local git remote, "
-    "e.g. org=Scaled-Tech-Consulting repo=Gitea-Tools."
+    "Pass explicit org= and repo= matching the local git remote on tools that "
+    "accept those parameters (including mcp_get_control_plane_guide), "
+    "e.g. org=Scaled-Tech-Consulting repo=Gitea-Tools. "
+    "Do not pass org/repo to tools whose schema does not list them."
 )
 
+# https://host/org/repo.git  or  git@host:org/repo.git
+_REMOTE_URL_SLUG_RE = re.compile(
+    r"(?:[:/])(?P<org>[^/]+)/(?P<repo>[^/]+?)(?:\.git)?/*$"
+)
+
+
+def parse_org_repo_from_remote_url(remote_url: str | None) -> tuple[str, str] | None:
+    """Best-effort parse of org/repo from a git remote URL."""
+    url = (remote_url or "").strip()
+    if not url:
+        return None
+    match = _REMOTE_URL_SLUG_RE.search(url)
+    if not match:
+        return None
+    org = (match.group("org") or "").strip()
+    repo = (match.group("repo") or "").strip()
+    if not org or not repo:
+        return None
+    return org, repo
+
 
 def assess_remote_repo_match(
     *,

review_final_report_schema.py

@@ -25,8 +25,13 @@ _REVIEWED_HEAD_RE = re.compile(
     r"(?:pinned reviewed head|reviewed head sha)\s*:\s*([0-9a-f]{7,40})",
     re.IGNORECASE,
 )
+# #698: validation pass proof appears in several legitimate shapes —
+# "Validation: pass", "Validation: focused 50 passed; full 2665 passed",
+# or structured "validation_status: pass". Accept pass evidence anywhere in
+# the Validation field's value, not only as its first token.
 _VALIDATION_PASS_RE = re.compile(
-    r"validation\s*:\s*(?:pass|passed|strong|ok|green)",
+    r"validation(?:_status)?\s*:[^\n]{0,300}?"
+    r"(?:\bpass(?:ed)?\b|\bstrong\b|\bok\b|\bgreen\b|\d+\s+passed)",
     re.IGNORECASE,
 )
 _MERGED_CLAIM_RE = re.compile(

review_merge_state_machine.py

@@ -93,8 +93,16 @@ def assess_workflow_blockers(
     capability_blocked: bool = False,
     mcp_reconnect_failed: bool = False,
     stale_capability_state: bool = False,
+    live_namespace_broken: bool = False,
 ) -> dict[str, Any]:
-    """Return hard blockers that forbid all PR queue work (#290 AC3)."""
+    """Return hard blockers that forbid all PR queue work (#290 AC3).
+
+    ``live_namespace_broken`` fails the merge/review path closed when the live
+    MCP namespace call path is unusable (e.g. ``client is closing: EOF``) even
+    though the tool is registered in FastMCP (#543 AC5). Feed it the
+    ``blocks_merge_workflow`` verdict from
+    ``mcp_namespace_health.classify_namespace_probe``.
+    """
     reasons: list[str] = []
     if infra_stop:
         reasons.append("infra_stop is active; PR selection/review/merge is forbidden")
@@ -104,6 +112,12 @@ def assess_workflow_blockers(
         reasons.append("MCP reconnect failed; stale session state cannot be reused")
     if stale_capability_state:
         reasons.append("stale MCP capability state detected after reconnect failure")
+    if live_namespace_broken:
+        reasons.append(
+            "live MCP namespace call path is broken (registered in FastMCP but "
+            "not callable through the namespace); repair the namespace before "
+            "review/merge"
+        )
     return {
         "block": bool(reasons),
         "reasons": reasons,
@@ -118,16 +132,19 @@ def assess_state_advancement(
     state_completion: dict[str, bool] | None,
     *,
     target_state: str,
-    infra_stop: bool = False,
-    capability_blocked: bool = False,
+    **blocker_kwargs,
 ) -> dict[str, Any]:
-    """Fail closed when *target_state* is requested before upstream gates pass."""
+    """Fail closed when *target_state* is requested before upstream gates pass.
+
+    Forwards every blocker flag (``infra_stop``, ``capability_blocked``,
+    ``mcp_reconnect_failed``, ``stale_capability_state``,
+    ``live_namespace_broken``) to :func:`assess_workflow_blockers` so
+    ``can_approve``/``can_merge`` honor the full blocker set, not just
+    infra_stop/capability_blocked (#543 AC5).
+    """
     completion = dict(state_completion or {})
     target = _clean(target_state).upper()
-    blockers = assess_workflow_blockers(
-        infra_stop=infra_stop,
-        capability_blocked=capability_blocked,
-    )
+    blockers = assess_workflow_blockers(**blocker_kwargs)
     reasons = list(blockers["reasons"])
 
     try:

review_proofs.py

@@ -858,9 +858,16 @@ _WALKTHROUGH_ARTIFACT_RE = re.compile(r"walkthrough\.md", re.I)
 
 
 def _performed_file_mutations(action_log: list[dict] | None) -> list[dict]:
-    """Return performed local file mutations, excluding gated rejections."""
+    """Return performed local file mutations, excluding gated rejections.
+
+    Non-dict entries (malformed JSON, LLM mistakes) are ignored instead of
+    raising ``AttributeError`` (#698): a malformed ledger entry can never be
+    authoritative mutation evidence.
+    """
     performed: list[dict] = []
     for entry in action_log or []:
+        if not isinstance(entry, dict):
+            continue
         if entry.get("gated_rejected") or entry.get("performed") is False:
             continue
         action = (entry.get("action") or "").strip().lower()
@@ -2228,20 +2235,39 @@ HANDOFF_REVIEW_MUTATION_FIELDS = (
 )
 
 HANDOFF_ROLE_FIELDS = {
+    # #698: the review/merger required-field sets must stay aligned with the
+    # canonical schema (skills/llm-project-workflow/schemas/
+    # review-merge-final-report.md). The schema explicitly FORBIDS the legacy
+    # fields 'Pinned reviewed head', 'Scratch worktree used', and 'Workspace
+    # mutations' — a validator must never demand a field the schema bans.
     "review": (
         ("Selected PR", ("selected pr",)),
         ("Reviewer eligibility", ("reviewer eligibility", "eligibility")),
-        ("Pinned reviewed head", ("pinned reviewed head", "pinned head")),
-        ("Worktree path", ("worktree path", "starting worktree path")),
-        ("Worktree dirty", ("worktree dirty", "whether worktree was dirty")),
-        ("Scratch worktree used", ("scratch worktree used", "scratch clone used",
-                                   "scratch worktree")),
+        ("Reviewed head SHA", ("reviewed head sha", "candidate head sha")),
+        ("Review worktree path", ("review worktree path", "worktree path",
+                                  "starting worktree path")),
+        ("Review worktree dirty", ("review worktree dirty", "worktree dirty",
+                                   "whether worktree was dirty")),
         ("Unrelated local mutations", ("unrelated local mutations",
-                                       "unrelated files modified")),
+                                       "unrelated files modified",
+                                       "file edits by reviewer")),
         ("Review decision", ("review decision", "decision")),
         ("Merge result", ("merge result",)),
         ("Linked issue status", ("linked issue status", "linked issue")),
         ("Cleanup status", ("cleanup status", "cleanup")),
+        ("Safe next action", ("safe next action", "next")),
+    ) + HANDOFF_REVIEW_MUTATION_FIELDS,
+    "merger": (
+        ("Selected PR", ("selected pr",)),
+        ("Reviewed head SHA", ("reviewed head sha", "candidate head sha")),
+        ("Active profile", ("active profile",)),
+        ("Role kind", ("role kind",)),
+        ("Merge capability source", ("merge capability source",)),
+        ("Explicit operator authorization", ("explicit operator authorization", "operator authorization", "authorization")),
+        ("Expected head SHA", ("expected head sha", "expected head")),
+        ("Approval at current head", ("approval at current head", "approval")),
+        ("Merge result", ("merge result",)),
+        ("Cleanup status", ("cleanup status", "cleanup")),
     ) + HANDOFF_REVIEW_MUTATION_FIELDS,
     "author": (
         ("Selected issue", ("selected issue",)),
@@ -2417,13 +2443,26 @@ def assess_controller_handoff(report_text, role=None, local_edits=False):
             field for field in required
             if field[0] not in ("Workspace mutations", "Mutations")
         ]
-    if role == "review":
-        # Issue #320: reviewer handoffs use the precise mutation categories
+    if role in ("review", "merger"):
+        # Issue #320: reviewer and merger handoffs use the precise mutation categories
         # in HANDOFF_REVIEW_MUTATION_FIELDS instead of the legacy ambiguous
         # "Workspace mutations" field, which is rejected below.
+        # Issue #698: the canonical review-merge schema has no 'Mutations',
+        # 'Next', 'Issue/PR', 'Branch/SHA', or 'Files changed' fields — their
+        # content lives in the precise mutation categories, 'Safe next
+        # action', 'Selected PR'/'Linked issue', head-SHA fields, and 'Files
+        # reviewed'. Requiring the legacy names rejects canonical reports.
+        _non_canonical_for_review = {
+            "Workspace mutations",
+            "Mutations",
+            "Next",
+            "Issue/PR",
+            "Branch/SHA",
+            "Files changed",
+        }
         required = [
             field for field in required
-            if field[0] != "Workspace mutations"
+            if field[0] not in _non_canonical_for_review
         ]
         if any(label.startswith("workspace mutations") for label in labels):
             return {
@@ -2431,7 +2470,7 @@ def assess_controller_handoff(report_text, role=None, local_edits=False):
                 "downgraded": True,
                 "missing_fields": [],
                 "reasons": [
-                    "review handoff must not include legacy "
+                    "review or merger handoff must not include legacy "
                     "'Workspace mutations' field; report the precise "
                     "mutation categories instead (issue #320)"
                 ],
@@ -4830,6 +4869,22 @@ RECONCILER_CLOSE_REPORT_FIELDS = (
     "no review/merge confirmation",
 )
 
+RECONCILER_SUPERSESSION_REPORT_FIELDS = (
+    "identity/profile",
+    "supersession close capability proof",
+    "target PR live state",
+    "target PR independent-work proof",
+    "superseding PR merged state",
+    "superseding merge commit SHA",
+    "target branch SHA",
+    "superseding ancestry proof",
+    "canonical close comment",
+    "linked issue status",
+    "PR close result",
+    "issue close result",
+    "no review/merge confirmation",
+)
+
 
 def assess_reconciler_close_gate(
     *,
@@ -4976,6 +5031,157 @@ def assess_reconciler_close_gate(
     }
 
 
+def assess_reconciler_supersession_close_gate(
+    *,
+    target_pr_number: int | None,
+    target_pr_state: str | None,
+    target_pr_mergeable: bool | None,
+    target_independently_required: bool | None,
+    superseding_pr_number: int | None,
+    superseding_pr_state: str | None,
+    superseding_pr_merged: bool,
+    superseding_merge_commit_sha: str | None,
+    superseding_head_sha: str | None,
+    target_branch: str | None,
+    target_branch_sha: str | None,
+    superseding_head_is_ancestor_of_target: bool | None,
+    canonical_comment_valid: bool,
+    canonical_comment_mentions_superseding_pr: bool,
+    canonical_comment_mentions_merge_commit: bool,
+    close_pr_capability: bool,
+    close_issue_capability: bool = False,
+    target_issue_state: str | None = None,
+    issue_satisfied_by_superseding: bool = False,
+) -> dict:
+    """Gate reconciler closure of PRs/issues superseded by a merged PR.
+
+    This is stricter than already-landed cleanup: the target PR may be closed
+    only after a named superseding PR is live-verified merged, its head is on a
+    freshly fetched target branch, the target PR is proven not independently
+    required, and a canonical close comment cites the superseding PR and merge
+    commit.
+    """
+    reasons: list[str] = []
+    if not isinstance(target_pr_number, int) or target_pr_number <= 0:
+        reasons.append("target PR number missing or invalid (#525)")
+    if not isinstance(superseding_pr_number, int) or superseding_pr_number <= 0:
+        reasons.append("superseding PR number missing or invalid (#525)")
+    if target_pr_number and superseding_pr_number and (
+        target_pr_number == superseding_pr_number
+    ):
+        reasons.append("target PR and superseding PR must differ (#525)")
+    if (target_pr_state or "").strip().lower() != "open":
+        reasons.append("target PR must be live-verified open before close (#525)")
+    if target_pr_mergeable is True:
+        reasons.append(
+            "target PR is still mergeable; prove it is superseded before close (#525)"
+        )
+    if target_independently_required is not False:
+        reasons.append(
+            "target PR independent-work proof missing; close requires explicit "
+            "proof that no required work remains (#525)"
+        )
+    if (superseding_pr_state or "").strip().lower() != "closed":
+        reasons.append("superseding PR must be live-verified closed (#525)")
+    if superseding_pr_merged is not True:
+        reasons.append("superseding PR must be live-verified merged (#525)")
+    if not _FULL_SHA.match((superseding_merge_commit_sha or "").strip()):
+        reasons.append("superseding merge commit SHA missing or invalid (#525)")
+    if not _FULL_SHA.match((superseding_head_sha or "").strip()):
+        reasons.append("superseding head SHA missing or invalid (#525)")
+    if not (target_branch or "").strip():
+        reasons.append("target branch missing (#525)")
+    if not _FULL_SHA.match((target_branch_sha or "").strip()):
+        reasons.append("target branch SHA missing or invalid (#525)")
+    if superseding_head_is_ancestor_of_target is None:
+        reasons.append("superseding ancestry proof not checked (#525)")
+    if not canonical_comment_valid:
+        reasons.append("canonical close comment failed validation (#525)")
+    if not canonical_comment_mentions_superseding_pr:
+        reasons.append("canonical comment must cite superseding PR (#525)")
+    if not canonical_comment_mentions_merge_commit:
+        reasons.append("canonical comment must cite merge commit SHA (#525)")
+
+    never_allowed = {
+        "review_allowed": False,
+        "approve_allowed": False,
+        "request_changes_allowed": False,
+        "merge_allowed": False,
+    }
+
+    if reasons:
+        return {
+            "outcome": "GATE_NOT_PROVEN",
+            "pr_close_allowed": False,
+            "issue_close_allowed": False,
+            "reasons": reasons,
+            "required_report_fields": RECONCILER_SUPERSESSION_REPORT_FIELDS,
+            "safe_next_action": (
+                "prove superseding PR state, target branch ancestry, target "
+                "supersession, and canonical comment before closing"
+            ),
+            **never_allowed,
+        }
+
+    if superseding_head_is_ancestor_of_target is False:
+        return {
+            "outcome": "SUPERSEDING_PR_NOT_ON_TARGET",
+            "pr_close_allowed": False,
+            "issue_close_allowed": False,
+            "reasons": [
+                f"superseding PR #{superseding_pr_number} head is not an "
+                f"ancestor of {target_branch}; supersession close denied (#525)"
+            ],
+            "required_report_fields": RECONCILER_SUPERSESSION_REPORT_FIELDS,
+            "safe_next_action": "re-fetch target branch and re-check ancestry",
+            **never_allowed,
+        }
+
+    if close_pr_capability is not True:
+        return {
+            "outcome": "RECOVERY_HANDOFF_REQUIRED",
+            "pr_close_allowed": False,
+            "issue_close_allowed": False,
+            "reasons": [
+                "exact gitea.pr.close capability not proven; produce a "
+                "recovery handoff instead of closing (#525)"
+            ],
+            "required_report_fields": RECONCILER_SUPERSESSION_REPORT_FIELDS,
+            "safe_next_action": (
+                "launch a reconciler profile with gitea.pr.close and replay "
+                "the live-state proof"
+            ),
+            **never_allowed,
+        }
+
+    issue_close_allowed = (
+        (target_issue_state or "").strip().lower() == "open"
+        and issue_satisfied_by_superseding is True
+        and close_issue_capability is True
+    )
+    issue_reasons = []
+    if (target_issue_state or "").strip().lower() == "closed":
+        issue_reasons.append("linked issue already closed; no issue close attempted (#525)")
+    elif target_issue_state and not issue_close_allowed:
+        issue_reasons.append(
+            "issue close requires an open issue satisfied by the superseding "
+            "merged PR and exact gitea.issue.close capability (#525)"
+        )
+
+    return {
+        "outcome": "SUPERSESSION_CLOSE_ALLOWED",
+        "pr_close_allowed": True,
+        "issue_close_allowed": issue_close_allowed,
+        "reasons": issue_reasons,
+        "required_report_fields": RECONCILER_SUPERSESSION_REPORT_FIELDS,
+        "safe_next_action": (
+            "post the canonical comment, close the superseded PR, and close "
+            "the linked issue only when issue_close_allowed is true"
+        ),
+        **never_allowed,
+    }
+
+
 # ---------------------------------------------------------------------------
 # Identity disclosure (#305)
 # ---------------------------------------------------------------------------
@@ -5540,6 +5746,15 @@ def assess_already_landed_classification_report(report_text, **kwargs):
     return _assess(report_text, **kwargs)
 
 
+def assess_conflict_fix_classification_final_report(report_text, **kwargs):
+    """#522: require live PR head re-pin before conflict-fix classification."""
+    from conflict_fix_classification import (
+        assess_conflict_fix_classification_final_report as _assess,
+    )
+
+    return _assess(report_text, **kwargs)
+
+
 def assess_prior_blocker_skip_proof(report_text, **kwargs):
     """#318: require live blocker proof before skipping earlier open PRs."""
     from reviewer_blocker_skip import assess_prior_blocker_skip_proof as _assess

review_quarantine.py

@@ -0,0 +1,351 @@
+"""Controller quarantine of contaminated formal reviews (#695).
+
+Quarantine records are durable under the MCP session-state root and are only
+writable when the caller holds a **native** MCP transport runtime. Untrusted
+local scripts that import this module cannot establish a valid quarantine
+(writes fail closed). Live server-side gates (eligibility, merge, feedback)
+honor quarantine by review_id + PR + head.
+
+Forensic evidence (Gitea review objects, lease comments) is never deleted.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+from datetime import datetime, timezone
+from typing import Any
+
+import mcp_daemon_guard
+import mcp_session_state
+
+KIND_QUARANTINE = "review_quarantine"
+CONFIRMATION_PREFIX = "QUARANTINE CONTAMINATED REVIEW"
+
+
+def _now() -> str:
+    return datetime.now(timezone.utc).isoformat()
+
+
+def quarantine_state_path(
+    *,
+    remote: str,
+    org: str,
+    repo: str,
+    pr_number: int,
+    review_id: int,
+) -> str:
+    # Dedicated subdir under session state root (mode 0o700).
+    root = os.path.join(mcp_session_state.default_state_dir(), "quarantine")
+    os.makedirs(root, mode=0o700, exist_ok=True)
+    # Explicit multi-segment key so remote/org/repo/pr/review cannot collide.
+    segs = [
+        KIND_QUARANTINE,
+        mcp_session_state._sanitize_segment(remote),
+        mcp_session_state._sanitize_segment(org),
+        mcp_session_state._sanitize_segment(repo),
+        f"pr{int(pr_number)}",
+        f"rev{int(review_id)}",
+    ]
+    return os.path.join(root, "-".join(segs) + ".json")
+
+
+def build_quarantine_record(
+    *,
+    remote: str,
+    org: str,
+    repo: str,
+    pr_number: int,
+    review_id: int,
+    reviewed_head_sha: str,
+    reason: str,
+    actor_username: str | None,
+    profile_name: str | None,
+    incident_issue: int | None = None,
+    forensic_comment_ids: list[int] | None = None,
+) -> dict[str, Any]:
+    provenance = mcp_daemon_guard.mutation_provenance_fields()
+    return {
+        "kind": KIND_QUARANTINE,
+        "issue_ref": "#695",
+        "remote": remote,
+        "org": org,
+        "repo": repo,
+        "pr_number": int(pr_number),
+        "review_id": int(review_id),
+        "reviewed_head_sha": (reviewed_head_sha or "").strip().lower(),
+        "reason": (reason or "").strip(),
+        "actor_username": actor_username,
+        "profile_name": profile_name,
+        "incident_issue": incident_issue,
+        "forensic_comment_ids": list(forensic_comment_ids or []),
+        "created_at": _now(),
+        "native_provenance": provenance,
+        "merge_authorization": "void",
+        "retain_forensic_evidence": True,
+    }
+
+
+def assess_quarantine_write(
+    *,
+    confirmation: str,
+    pr_number: int,
+    review_id: int,
+    reason: str,
+    native_required: bool = True,
+) -> dict[str, Any]:
+    """Pure assessment of whether a quarantine write may proceed."""
+    reasons: list[str] = []
+    expected = f"{CONFIRMATION_PREFIX} {int(review_id)} PR {int(pr_number)}"
+    if (confirmation or "").strip() != expected:
+        reasons.append(
+            f"confirmation must equal exactly '{expected}' (fail closed, #695)"
+        )
+    if not (reason or "").strip():
+        reasons.append("quarantine reason is required (fail closed, #695)")
+    if native_required and not mcp_daemon_guard.is_native_mcp_transport():
+        if not mcp_daemon_guard.is_pytest_runtime():
+            reasons.append(
+                "quarantine write requires native MCP transport; untrusted "
+                "local code cannot create quarantine records (#695)"
+            )
+    return {
+        "allowed": not reasons,
+        "expected_confirmation": expected,
+        "reasons": reasons,
+    }
+
+
+def write_quarantine_record(record: dict[str, Any]) -> dict[str, Any]:
+    """Persist quarantine record; fail closed outside native/pytest runtime."""
+    if not mcp_daemon_guard.is_native_mcp_transport():
+        if not mcp_daemon_guard.is_pytest_runtime():
+            raise mcp_daemon_guard.UnsanctionedRuntimeError(
+                "quarantine write blocked: non-native runtime (#695)"
+            )
+    path = quarantine_state_path(
+        remote=str(record["remote"]),
+        org=str(record["org"]),
+        repo=str(record["repo"]),
+        pr_number=int(record["pr_number"]),
+        review_id=int(record["review_id"]),
+    )
+    # Refuse overwriting with weaker provenance from untrusted caller by
+    # requiring native fields present.
+    if not (record.get("native_provenance") or {}).get("native_mcp_transport"):
+        if not mcp_daemon_guard.is_pytest_runtime():
+            raise mcp_daemon_guard.UnsanctionedRuntimeError(
+                "quarantine record missing native provenance (#695)"
+            )
+    tmp = path + ".tmp"
+    data = json.dumps(record, indent=2, sort_keys=True)
+    with open(tmp, "w", encoding="utf-8") as fh:
+        fh.write(data)
+        fh.flush()
+        os.fsync(fh.fileno())
+    os.replace(tmp, path)
+    os.chmod(path, 0o600)
+    return {"path": path, "written": True, "record": record}
+
+
+def load_quarantine_record(
+    *,
+    remote: str,
+    org: str,
+    repo: str,
+    pr_number: int,
+    review_id: int,
+) -> dict[str, Any] | None:
+    path = quarantine_state_path(
+        remote=remote,
+        org=org,
+        repo=repo,
+        pr_number=pr_number,
+        review_id=review_id,
+    )
+    if not os.path.isfile(path):
+        return None
+    try:
+        with open(path, encoding="utf-8") as fh:
+            data = json.load(fh)
+    except (OSError, json.JSONDecodeError):
+        return None
+    if not isinstance(data, dict):
+        return None
+    return data
+
+
+def is_review_quarantined(
+    *,
+    remote: str,
+    org: str,
+    repo: str,
+    pr_number: int,
+    review_id: int | None,
+    reviewed_head_sha: str | None = None,
+) -> dict[str, Any]:
+    """Whether a formal review is quarantined for merge authorization (#695)."""
+    if review_id is None:
+        return {"quarantined": False, "record": None, "reasons": []}
+    rec = load_quarantine_record(
+        remote=remote,
+        org=org,
+        repo=repo,
+        pr_number=pr_number,
+        review_id=int(review_id),
+    )
+    if not rec:
+        return {"quarantined": False, "record": None, "reasons": []}
+    reasons = [
+        f"formal review_id {review_id} on PR #{pr_number} is quarantined "
+        f"(#695; incident issue {rec.get('incident_issue')}); "
+        "merge authorization is void; forensic evidence retained"
+    ]
+    want_head = (reviewed_head_sha or "").strip().lower()
+    rec_head = (rec.get("reviewed_head_sha") or "").strip().lower()
+    if want_head and rec_head and want_head != rec_head:
+        # Still quarantined by review_id; note head mismatch for operators.
+        reasons.append(
+            f"quarantine head {rec_head[:12]}… differs from assessed head "
+            f"{want_head[:12]}… (still void by review_id)"
+        )
+    return {"quarantined": True, "record": rec, "reasons": reasons}
+
+
+def filter_approvals_for_merge(
+    *,
+    remote: str,
+    org: str,
+    repo: str,
+    pr_number: int,
+    current_head_sha: str | None,
+    reviews: list[dict],
+) -> dict[str, Any]:
+    """Split reviews into merge-usable vs quarantined contaminated approvals."""
+    current = (current_head_sha or "").strip().lower()
+    usable: list[dict] = []
+    quarantined: list[dict] = []
+    for rev in reviews or []:
+        if not isinstance(rev, dict):
+            continue
+        verdict = (rev.get("verdict") or rev.get("state") or "").upper()
+        if verdict not in {"APPROVED", "APPROVE"}:
+            continue
+        if rev.get("dismissed"):
+            continue
+        rid = rev.get("review_id") or rev.get("id")
+        head = (
+            rev.get("reviewed_head_sha")
+            or rev.get("commit_id")
+            or rev.get("head_sha")
+            or ""
+        ).strip().lower()
+        q = is_review_quarantined(
+            remote=remote,
+            org=org,
+            repo=repo,
+            pr_number=pr_number,
+            review_id=int(rid) if rid is not None else None,
+            reviewed_head_sha=head or current,
+        )
+        entry = {**rev, "review_id": rid, "reviewed_head_sha": head}
+        if q["quarantined"]:
+            entry["quarantined"] = True
+            entry["quarantine_reasons"] = q["reasons"]
+            quarantined.append(entry)
+        else:
+            entry["quarantined"] = False
+            usable.append(entry)
+    usable_at_head = [
+        e
+        for e in usable
+        if current and (e.get("reviewed_head_sha") or "") == current
+    ]
+    return {
+        "usable_approvals": usable,
+        "usable_approvals_at_current_head": usable_at_head,
+        "quarantined_approvals": quarantined,
+        "approval_visible_for_merge": bool(usable_at_head),
+        "has_quarantined_approval_at_head": any(
+            current
+            and (e.get("reviewed_head_sha") or "") == current
+            for e in quarantined
+        ),
+    }
+
+
+def format_quarantine_audit_comment(record: dict[str, Any]) -> str:
+    """Append-only forensic audit comment body (does not delete evidence)."""
+    lines = [
+        "## Contaminated formal review quarantine (#695)",
+        "",
+        "Status: **QUARANTINED — merge authorization VOID**",
+        "",
+        f"- review_id: `{record.get('review_id')}`",
+        f"- pr: `#{record.get('pr_number')}`",
+        f"- reviewed_head_sha: `{record.get('reviewed_head_sha')}`",
+        f"- actor: `{record.get('actor_username')}`",
+        f"- profile: `{record.get('profile_name')}`",
+        f"- incident_issue: `#{record.get('incident_issue')}`",
+        f"- reason: {record.get('reason')}",
+        f"- created_at: `{record.get('created_at')}`",
+        f"- native_transport: "
+        f"`{(record.get('native_provenance') or {}).get('native_mcp_transport')}`",
+        f"- native_token_fingerprint: "
+        f"`{(record.get('native_provenance') or {}).get('native_token_fingerprint')}`",
+        f"- forensic_comment_ids retained: "
+        f"`{record.get('forensic_comment_ids')}`",
+        "",
+        "This record does **not** delete Gitea reviews or historical comments. "
+        "Fresh native-MCP re-review is required before merge.",
+    ]
+    return "\n".join(lines)
+
+
+def assess_untrusted_canonical_approval_claim(body: str) -> list[str]:
+    """Reasons to reject canonical comments that claim approved/merge-ready.
+
+    Used when the comment asserts approval without native review proof (#695 AC7).
+    False "official workflow" claims that cite offline/import paths are always
+    rejected even when a NATIVE_REVIEW_PROOF line is present.
+    """
+    text = body or ""
+    lower = text.lower()
+    claims_approved = (
+        "state:\napproved" in lower
+        or "state: approved" in lower
+        or "who_is_next:\nmerger" in lower
+        or "who_is_next: merger" in lower
+        or "merge_ready: true" in lower
+        or "merge_ready:\ntrue" in lower
+        or "ready-to-merge" in lower
+    )
+    if not claims_approved:
+        return []
+    # Reject explicit offline/import "official workflow" spoofing (#695 AC9).
+    offline_spoof = (
+        "offline_mcp" in lower
+        or "offline import" in lower
+        or "direct import" in lower
+        or "import gitea_mcp_server" in lower
+        or "run_quarantine.py" in lower
+        or "offline_mcp_helper" in lower
+        or "offline_mcp_runner" in lower
+    )
+    has_proof = (
+        "NATIVE_REVIEW_PROOF:" in text
+        or "native_review_proof:" in lower
+    )
+    if offline_spoof:
+        return [
+            "canonical approval claim cites offline/import/helper path; "
+            "NATIVE_REVIEW_PROOF rejected (#695 AC7/AC9); stop after native "
+            "MCP failure — do not construct offline mutation fallbacks"
+        ]
+    if has_proof:
+        return []
+    return [
+        "canonical comment claims approved/merge-ready state without "
+        "NATIVE_REVIEW_PROOF from a native MCP review mutation (#695 AC7); "
+        "contaminated or untrusted approvals cannot certify merger handoff"
+    ]

reviewer_handoff_consistency.py

@@ -0,0 +1,186 @@
+"""Reviewer handoff consistency validation (#501).
+
+Detects contradictions between narrative claims and mutation ledger fields in
+reviewer/final-review controller handoffs. Pure validation only — does not post
+comments or mutate Gitea state.
+"""
+
+from __future__ import annotations
+
+import re
+
+_HANDOFF_FIELD_RE = re.compile(
+    r"^\s*[-*]\s*([^:]+):\s*(.*)$",
+    re.MULTILINE | re.IGNORECASE,
+)
+
+_MERGE_NARRATIVE_RE = re.compile(
+    r"\b(?:merged\s+pr\s*#|pr\s+#\d+\s+(?:was\s+)?merged|"
+    r"merge\s+result\s*:\s*merged|successfully\s+merged|"
+    r"terminal\s+mutation\s+budget.{0,80}\bmerge)\b",
+    re.IGNORECASE | re.DOTALL,
+)
+_REVIEW_SUBMIT_BLOCKED_RE = re.compile(
+    r"\b(?:review\s+submission\s+blocked|submit_pr_review\s+(?:failed|blocked)|"
+    r"gitea_submit_pr_review\s+(?:failed|blocked|not\s+(?:attempted|performed))|"
+    r"review\s+not\s+submitted|could\s+not\s+submit\s+review)\b",
+    re.IGNORECASE,
+)
+_FINAL_DECISION_MARKED_RE = re.compile(
+    r"\b(?:gitea_mark_final_review_decision|final\s+review\s+decision\s+marked|"
+    r"server-side\s+final\s+decision\s+marked|marked\s+final\s+review\s+decision)\b",
+    re.IGNORECASE,
+)
+_TERMINAL_BUDGET_CONSUMED_RE = re.compile(
+    r"\b(?:terminal\s+mutation\s+budget\s+(?:already\s+)?consumed|"
+    r"single[- ]terminal\s+mutation\s+(?:already\s+)?consumed|"
+    r"mutation\s+budget\s+exhausted)\b",
+    re.IGNORECASE,
+)
+_LEASE_NARRATIVE_RE = re.compile(
+    r"\b(?:reviewer\s+lease\s+acquired|acquired\s+reviewer\s+pr\s+lease|"
+    r"gitea_acquire_reviewer_pr_lease)\b",
+    re.IGNORECASE,
+)
+_REJECTED_MUTATION_RE = re.compile(
+    r"\b(?:mutation\s+rejected|tool\s+failed\s+closed|blocked\s+before\s+mutation|"
+    r"no\s+server-side\s+state\s+changed)\b",
+    re.IGNORECASE,
+)
+_MERGE_IN_LEDGER_RE = re.compile(r"\bmerge\b", re.IGNORECASE)
+_REVIEW_SUBMITTED_IN_LEDGER_RE = re.compile(
+    r"\b(?:submitted|approve|request_changes|review\s+submitted)\b",
+    re.IGNORECASE,
+)
+_NONE_VALUE_RE = re.compile(r"^\s*(?:none|not\s+attempted|n/a)\s*$", re.IGNORECASE)
+
+_BLOCKED_REVIEW_PROOF_FIELDS = (
+    "tool called",
+    "mutation attempted",
+    "mutation rejected",
+    "no server-side state changed",
+)
+
+
+def render_blocked_review_handoff_template() -> str:
+    """Return a corrected handoff template for blocked review submissions."""
+    return """## Blocked Review Submission Handoff
+
+- Tool called: gitea_submit_pr_review
+- Mutation attempted: yes
+- Mutation rejected: yes
+- No server-side state changed: confirmed
+- Proof/source: <paste exact tool error or gate reason>
+- Prior terminal mutation that consumed budget: <name exact prior mutation or none>
+- Review decision (local intent): <approve | request_changes | comment only>
+- Server-side final decision marked: <yes with tool proof | no>
+- Gitea review submitted: no
+- Gitea review blocked because: <exact gate/error>
+- Review mutations: none (submission blocked)
+- MCP/Gitea mutations: <list only mutations that actually occurred>
+- Safe next action: rewrite handoff with consistent ledger before posting
+"""
+
+
+def _handoff_fields(text: str | None) -> dict[str, str]:
+    fields: dict[str, str] = {}
+    for match in _HANDOFF_FIELD_RE.finditer(text or ""):
+        key = match.group(1).strip().lower()
+        value = match.group(2).strip()
+        fields[key] = value
+    return fields
+
+
+def _is_none_value(value: str | None) -> bool:
+    return bool(_NONE_VALUE_RE.match(value or ""))
+
+
+def _ledger_mentions_merge(*values: str | None) -> bool:
+    blob = " ".join(v for v in values if v)
+    return bool(blob) and bool(_MERGE_IN_LEDGER_RE.search(blob))
+
+
+def _ledger_mentions_review_submission(*values: str | None) -> bool:
+    blob = " ".join(v for v in values if v)
+    return bool(blob) and bool(_REVIEW_SUBMITTED_IN_LEDGER_RE.search(blob))
+
+
+def assess_reviewer_handoff_consistency(report_text: str | None) -> dict:
+    """Validate reviewer handoff narrative against mutation ledger fields."""
+    text = report_text or ""
+    fields = _handoff_fields(text)
+    reasons: list[str] = []
+
+    review_mutations = fields.get("review mutations", "")
+    merge_mutations = fields.get("merge mutations", "")
+    mcp_mutations = fields.get("mcp/gitea mutations", "")
+    review_decision = fields.get("review decision", "")
+
+    if _MERGE_NARRATIVE_RE.search(text) and not _ledger_mentions_merge(
+        merge_mutations, mcp_mutations, review_mutations
+    ):
+        reasons.append(
+            "narrative claims a merge occurred but mutation ledger omits merge"
+        )
+
+    if _TERMINAL_BUDGET_CONSUMED_RE.search(text):
+        if _ledger_mentions_merge(merge_mutations, mcp_mutations):
+            pass
+        elif _LEASE_NARRATIVE_RE.search(text) and not _ledger_mentions_merge(
+            merge_mutations, mcp_mutations
+        ):
+            reasons.append(
+                "terminal mutation budget attributed to merge but ledger only "
+                "shows lease or non-merge activity"
+            )
+        elif not _ledger_mentions_merge(merge_mutations, mcp_mutations):
+            reasons.append(
+                "terminal mutation budget consumed claim must identify the "
+                "exact prior mutation in the ledger"
+            )
+
+    if _REVIEW_SUBMIT_BLOCKED_RE.search(text) and _FINAL_DECISION_MARKED_RE.search(text):
+        reasons.append(
+            "handoff claims review submission blocked but also claims "
+            "server-side final decision was marked"
+        )
+
+    lease_claimed = _LEASE_NARRATIVE_RE.search(text) or (
+        not _is_none_value(mcp_mutations)
+        and "lease" in mcp_mutations.lower()
+    )
+    if lease_claimed and _is_none_value(review_decision):
+        reasons.append(
+            "reviewer lease acquired but Review decision field is missing or none"
+        )
+
+    if _REVIEW_SUBMIT_BLOCKED_RE.search(text) or _REJECTED_MUTATION_RE.search(text):
+        lower = text.lower()
+        missing_proof = [
+            field
+            for field in _BLOCKED_REVIEW_PROOF_FIELDS
+            if field not in lower
+        ]
+        if missing_proof:
+            reasons.append(
+                "blocked/rejected mutation claim missing proof fields: "
+                + ", ".join(missing_proof)
+            )
+
+    if (
+        _FINAL_DECISION_MARKED_RE.search(text)
+        and _is_none_value(review_mutations)
+        and not _ledger_mentions_review_submission(mcp_mutations)
+        and not _REVIEW_SUBMIT_BLOCKED_RE.search(text)
+    ):
+        reasons.append(
+            "final review decision marked but Review mutations ledger is empty"
+        )
+
+    proven = not reasons
+    return {
+        "proven": proven,
+        "block": not proven,
+        "reasons": reasons,
+        "fields": fields,
+    }

reviewer_pr_lease.py

@@ -190,18 +190,28 @@ def find_active_reviewer_lease(
     pr_number: int,
     now: datetime | None = None,
 ) -> dict[str, Any] | None:
-    """Newest non-terminal, unexpired lease for *pr_number*."""
+    """Return the newest unexpired non-terminal lease for *pr_number*.
+
+    Append-only lease markers form a ledger: the **newest** marker is
+    authoritative. A later terminal phase (``released`` / ``done`` /
+    ``blocked``) ends the lease even when older ``claimed`` markers remain
+    on the thread (#577). Skipping only terminal markers and walking older
+    claims incorrectly re-arms a lease after a successful release.
+    """
     now = now or datetime.now(timezone.utc)
-    for lease in reversed(_lease_entries(comments, pr_number=pr_number)):
-        phase = (lease.get("phase") or "").strip().lower()
-        if phase in _TERMINAL_PHASES:
-            continue
-        if _lease_expired(lease, now=now):
-            continue
-        if phase in _ACTIVE_PHASES or phase:
-            lease = dict(lease)
-            lease["freshness"] = classify_lease_freshness(lease, now=now)
-            return lease
+    entries = list(reversed(_lease_entries(comments, pr_number=pr_number)))
+    if not entries:
+        return None
+    newest = entries[0]
+    phase = (newest.get("phase") or "").strip().lower()
+    if phase in _TERMINAL_PHASES:
+        return None
+    if _lease_expired(newest, now=now):
+        return None
+    if phase in _ACTIVE_PHASES or phase:
+        lease = dict(newest)
+        lease["freshness"] = classify_lease_freshness(lease, now=now)
+        return lease
     return None
 
 
@@ -512,4 +522,855 @@ def assess_lease_inventory(
         "active_review_leases": active,
         "stale_review_leases": stale,
         "reclaimable_review_leases": reclaimable,
-    }
+    }
+
+# Canonical next-action vocabulary for reviewer lease handoff (#599, #691).
+NEXT_ACTION_ACQUIRE = "acquire"
+NEXT_ACTION_WAIT = "wait"
+NEXT_ACTION_RESUME_EXACT_OWNER_SESSION = "resume_exact_owner_session"
+NEXT_ACTION_RELEASE_EXPIRED_LEASE = "release_expired_lease"
+NEXT_ACTION_OPERATOR_AUTHORIZED_CLEANUP = "operator_authorized_cleanup"
+NEXT_ACTION_REPAIR_WORKTREE_BINDING = "repair_worktree_binding"
+NEXT_ACTION_CLEANUP_OBSOLETE_LEASE = "cleanup_obsolete_reviewer_comment_lease"
+
+CLEANUP_OBSOLETE_LEASE_TOOL = "gitea_cleanup_obsolete_reviewer_comment_lease"
+CLEANUP_OBSOLETE_LEASE_CONFIRMATION_PREFIX = "CLEANUP OBSOLETE REVIEWER LEASE "
+
+# Formal terminal review verdicts that complete a leased-head workflow (#691).
+_TERMINAL_REVIEW_VERDICTS = frozenset({
+    "APPROVED",
+    "REQUEST_CHANGES",
+    "approved",
+    "request_changes",
+    "REQUESTCHANGES",
+})
+
+_HANDOFF_CLASSIFICATIONS = frozenset({
+    "no_lease",
+    "own_active",
+    "own_expired",
+    "foreign_active",
+    "foreign_reclaimable",
+    "foreign_expired",
+    "foreign_active_current_head",
+    "foreign_expired_current_head",
+    "foreign_completed_superseded_head",
+    "foreign_expired_superseded_head",
+    "orphaned_owner_missing",
+    "ambiguous_conflicting_evidence",
+    "instructed_lease_missing_with_replacement",
+    "worktree_binding_mismatch",
+})
+
+
+def _norm_path(value: str | None) -> str:
+    text = (value or "").strip()
+    if not text:
+        return ""
+    return os.path.normpath(text.rstrip("/"))
+
+
+def _normalize_verdict(value: str | None) -> str:
+    text = (value or "").strip().upper().replace("-", "_").replace(" ", "_")
+    if text in {"REQUESTCHANGES", "REQUEST_CHANGES", "CHANGES_REQUESTED"}:
+        return "REQUEST_CHANGES"
+    if text in {"APPROVED", "APPROVE"}:
+        return "APPROVED"
+    return text
+
+
+def formal_terminal_review_for_head(
+    formal_reviews: list[dict] | None,
+    *,
+    leased_head: str | None,
+) -> dict[str, Any] | None:
+    """Return the latest undismissed terminal formal review for *leased_head*."""
+    head = _normalize_sha(leased_head)
+    if not head:
+        return None
+    matches: list[dict[str, Any]] = []
+    for review in formal_reviews or []:
+        if review.get("dismissed"):
+            continue
+        verdict = _normalize_verdict(
+            review.get("verdict") or review.get("state") or review.get("body_state")
+        )
+        if verdict not in {"APPROVED", "REQUEST_CHANGES"}:
+            continue
+        rhead = _normalize_sha(
+            review.get("reviewed_head_sha")
+            or review.get("commit_id")
+            or review.get("head_sha")
+        )
+        if rhead != head:
+            continue
+        matches.append({**review, "verdict": verdict, "reviewed_head_sha": rhead})
+    if not matches:
+        return None
+    return matches[-1]
+
+
+def find_newest_nonterminal_lease(
+    comments: list[dict],
+    *,
+    pr_number: int,
+    now: datetime | None = None,
+    include_expired: bool = True,
+) -> dict[str, Any] | None:
+    """Newest non-terminal lease marker, optionally including expired ones (#691).
+
+    Unlike ``find_active_reviewer_lease``, this retains expired non-terminal
+    markers so diagnosis/cleanup can still name the obsolete lease after
+    ``expires_at`` (comment-backed ledger; no control-plane ``lease_id``).
+    """
+    now = now or datetime.now(timezone.utc)
+    entries = list(reversed(_lease_entries(comments, pr_number=pr_number)))
+    for entry in entries:
+        phase = (entry.get("phase") or "").strip().lower()
+        if phase in _TERMINAL_PHASES:
+            # Newest terminal ends the ledger chain for active acquisition, but
+            # an older non-terminal is not "active". Stop at newest marker.
+            return None
+        expired = _lease_expired(entry, now=now)
+        if expired and not include_expired:
+            return None
+        lease = dict(entry)
+        lease["freshness"] = classify_lease_freshness(lease, now=now)
+        lease["expired"] = expired
+        return lease
+    return None
+
+
+def cleanup_confirmation_for_pr(pr_number: int) -> str:
+    return f"{CLEANUP_OBSOLETE_LEASE_CONFIRMATION_PREFIX}{int(pr_number)}"
+
+
+def assess_obsolete_reviewer_comment_lease_cleanup(
+    comments: list[dict],
+    *,
+    pr_number: int,
+    current_head_sha: str | None,
+    formal_reviews: list[dict] | None = None,
+    repo: str | None = None,
+    expected_repo: str | None = None,
+    requesting_session_id: str | None = None,
+    controller_recovery_authorized: bool = False,
+    worktree_exists: bool | None = None,
+    worktree_clean: bool | None = None,
+    worktree_has_unpreserved_work: bool | None = None,
+    owner_process_alive: bool | None = None,
+    owner_pid_observed: int | None = None,
+    requesting_pid: int | None = None,
+    current_head_review_in_progress: bool = False,
+    expected_lease_comment_id: int | None = None,
+    expected_session_id: str | None = None,
+    expected_leased_head: str | None = None,
+    confirmation: str | None = None,
+    apply: bool = False,
+    now: datetime | None = None,
+) -> dict[str, Any]:
+    """Guarded cleanup eligibility for obsolete comment-backed reviewer leases (#691).
+
+    Cleanup is never transfer/adoption/repoint and never uses PID equality as
+    ownership. Eligible only when evidence proves the lease is past expiry and/or
+    pinned to a superseded head with a completed formal terminal review, and
+    every safety boundary holds.
+    """
+    now = now or datetime.now(timezone.utc)
+    reasons: list[str] = []
+    fail_closed_reasons: list[str] = []
+    current_head = _normalize_sha(current_head_sha)
+    req_session = (requesting_session_id or "").strip()
+
+    lease = find_newest_nonterminal_lease(
+        comments, pr_number=pr_number, now=now, include_expired=True
+    )
+    if not lease:
+        # Fall back to active finder (unexpired only) for report consistency.
+        lease = find_active_reviewer_lease(comments, pr_number=pr_number, now=now)
+
+    classification = "no_lease"
+    cleanup_allowed = False
+    release_body: str | None = None
+    audit_comment_body: str | None = None
+    terminal_review = None
+    leased_head = None
+    head_superseded = False
+    past_expiry = False
+    expires_parseable = True
+
+    if not lease:
+        fail_closed_reasons.append(
+            f"PR #{pr_number}: no non-terminal comment-backed reviewer lease to clean"
+        )
+        classification = "no_lease"
+    else:
+        leased_head = _normalize_sha(lease.get("candidate_head"))
+        expires_raw = lease.get("expires_at")
+        expires_at = _parse_timestamp(expires_raw)
+        if expires_raw and expires_at is None:
+            expires_parseable = False
+            fail_closed_reasons.append(
+                "lease expires_at cannot be parsed or trusted (fail closed)"
+            )
+        past_expiry = bool(expires_at and expires_at <= now)
+        freshness = lease.get("freshness") or classify_lease_freshness(lease, now=now)
+        owner_session = (lease.get("session_id") or "").strip()
+        is_requesting_owner = bool(
+            req_session and owner_session and req_session == owner_session
+        )
+
+        # Identity pins (repo / PR / session / head / comment) must match when
+        # the caller supplies expected evidence.
+        lease_repo = (lease.get("repo") or "").strip()
+        exp_repo = (expected_repo or repo or "").strip()
+        if exp_repo and lease_repo and exp_repo != lease_repo:
+            fail_closed_reasons.append(
+                f"repository identity mismatch: lease repo={lease_repo!r} "
+                f"expected={exp_repo!r}"
+            )
+        if lease.get("pr_number") not in (None, pr_number):
+            fail_closed_reasons.append(
+                f"PR identity mismatch: lease pr={lease.get('pr_number')} "
+                f"requested={pr_number}"
+            )
+        if expected_lease_comment_id is not None:
+            cid = lease.get("comment_id")
+            if cid is None or int(cid) != int(expected_lease_comment_id):
+                fail_closed_reasons.append(
+                    "lease comment_id does not match expected evidence "
+                    f"(lease={cid}, expected={expected_lease_comment_id})"
+                )
+        if expected_session_id:
+            if owner_session != expected_session_id.strip():
+                fail_closed_reasons.append(
+                    "lease session_id does not match expected evidence "
+                    f"(lease={owner_session}, expected={expected_session_id})"
+                )
+        if expected_leased_head:
+            exp_head = _normalize_sha(expected_leased_head)
+            if not exp_head or exp_head != leased_head:
+                fail_closed_reasons.append(
+                    "leased head does not match expected evidence "
+                    f"(lease={leased_head}, expected={exp_head})"
+                )
+
+        # PID equality is never ownership proof (#691).
+        if (
+            owner_pid_observed is not None
+            and requesting_pid is not None
+            and int(owner_pid_observed) == int(requesting_pid)
+        ):
+            reasons.append(
+                "observed PID equals requesting PID but PID equality is NOT "
+                "ownership proof; ignored for authorization"
+            )
+
+        if is_requesting_owner:
+            fail_closed_reasons.append(
+                "requesting session owns the lease; use "
+                "gitea_release_reviewer_pr_lease (owner path), not non-owner cleanup"
+            )
+
+        if current_head_review_in_progress:
+            fail_closed_reasons.append(
+                "a current-head review submission is in progress; cleanup denied"
+            )
+
+        if not current_head:
+            fail_closed_reasons.append(
+                "current PR head SHA missing or unparseable (fail closed)"
+            )
+        if not leased_head:
+            fail_closed_reasons.append(
+                "lease candidate_head missing or unparseable (fail closed)"
+            )
+
+        head_superseded = bool(
+            current_head and leased_head and current_head != leased_head
+        )
+        head_matches_current = bool(
+            current_head and leased_head and current_head == leased_head
+        )
+
+        terminal_review = formal_terminal_review_for_head(
+            formal_reviews, leased_head=leased_head
+        )
+        has_terminal = terminal_review is not None
+
+        # Worktree safety.
+        if worktree_has_unpreserved_work is True or worktree_clean is False:
+            fail_closed_reasons.append(
+                "lease worktree is dirty or has unpreserved work; cleanup denied"
+            )
+        if (
+            worktree_exists is True
+            and worktree_clean is None
+            and worktree_has_unpreserved_work is None
+        ):
+            # Unknown cleanliness when path exists — fail closed.
+            fail_closed_reasons.append(
+                "lease worktree exists but cleanliness is unknown (fail closed)"
+            )
+
+        # Classification matrix (#691).
+        if head_matches_current and freshness in {"active", "stale_warning"}:
+            classification = "foreign_active_current_head"
+            fail_closed_reasons.append(
+                "genuinely active foreign lease on the current PR head; "
+                "cleanup denied (fail closed)"
+            )
+        elif head_matches_current and past_expiry:
+            classification = "foreign_expired_current_head"
+            # Expired on current head: owner-missing / orphan path may apply.
+            if owner_process_alive is True:
+                fail_closed_reasons.append(
+                    "lease expired on current head but owner process still alive "
+                    "with plausible activity; cleanup denied"
+                )
+            elif worktree_clean is False:
+                fail_closed_reasons.append(
+                    "owner process absent but worktree dirty; cleanup denied"
+                )
+            elif owner_process_alive is False and (
+                worktree_clean is True or worktree_exists is False
+            ):
+                if controller_recovery_authorized:
+                    classification = "orphaned_owner_missing"
+                else:
+                    fail_closed_reasons.append(
+                        "controller/recovery capability not authorized for orphan cleanup"
+                    )
+            else:
+                fail_closed_reasons.append(
+                    "insufficient orphan evidence for expired current-head lease "
+                    "(need owner_process_alive=false and clean/absent worktree)"
+                )
+        elif head_superseded and has_terminal and past_expiry:
+            classification = "foreign_expired_superseded_head"
+        elif head_superseded and has_terminal and not past_expiry:
+            classification = "foreign_completed_superseded_head"
+        elif head_superseded and not has_terminal:
+            classification = "ambiguous_conflicting_evidence"
+            fail_closed_reasons.append(
+                "lease head is superseded but no formal terminal review exists "
+                "for the leased head (fail closed)"
+            )
+        elif not head_superseded and not past_expiry and freshness in {
+            "active", "stale_warning"
+        }:
+            classification = "foreign_active_current_head"
+            fail_closed_reasons.append(
+                "foreign lease still active on current head; wait or use owner release"
+            )
+        elif past_expiry and not head_superseded:
+            classification = "foreign_expired_current_head"
+        else:
+            classification = "ambiguous_conflicting_evidence"
+            fail_closed_reasons.append(
+                "lease evidence is ambiguous; cleanup denied (fail closed)"
+            )
+
+        # Recent owner progress on a non-superseded active lease blocks cleanup.
+        minutes = _minutes_since_activity(lease, now=now)
+        if (
+            head_matches_current
+            and freshness == "active"
+            and minutes is not None
+            and minutes < STALE_WARNING_MINUTES
+        ):
+            fail_closed_reasons.append(
+                "owner session has recent authenticated progress on current head; "
+                "cleanup denied"
+            )
+
+        # Authority + confirmation for apply.
+        if not controller_recovery_authorized:
+            if classification in {
+                "foreign_completed_superseded_head",
+                "foreign_expired_superseded_head",
+                "foreign_expired_current_head",
+                "orphaned_owner_missing",
+            }:
+                fail_closed_reasons.append(
+                    "controller/recovery capability required "
+                    "(controller_recovery_authorized=true)"
+                )
+
+        expected_conf = cleanup_confirmation_for_pr(pr_number)
+        if apply:
+            if (confirmation or "").strip() != expected_conf:
+                fail_closed_reasons.append(
+                    f"confirmation must equal exactly {expected_conf!r}"
+                )
+
+        # Superseded + terminal path does not require past_expiry (AC1).
+        eligible_class = classification in {
+            "foreign_completed_superseded_head",
+            "foreign_expired_superseded_head",
+            "orphaned_owner_missing",
+            "foreign_expired_current_head",
+        }
+        # foreign_expired_current_head needs orphan/clean worktree + authority.
+        if classification == "foreign_expired_current_head":
+            if worktree_clean is not True and worktree_exists is not False:
+                if "worktree" not in " ".join(fail_closed_reasons):
+                    fail_closed_reasons.append(
+                        "expired current-head cleanup requires proven-clean "
+                        "worktree or absent worktree"
+                    )
+            if owner_process_alive is True:
+                fail_closed_reasons.append(
+                    "owner process still alive on expired current-head lease"
+                )
+
+        cleanup_allowed = (
+            eligible_class
+            and expires_parseable
+            and not fail_closed_reasons
+            and not is_requesting_owner
+        )
+
+        if cleanup_allowed:
+            release_body = format_lease_body(
+                repo=lease.get("repo") or exp_repo or "",
+                pr_number=pr_number,
+                issue_number=lease.get("issue_number"),
+                reviewer_identity=lease.get("reviewer_identity") or "",
+                profile=lease.get("profile") or "unknown",
+                session_id=owner_session or "unknown",
+                worktree=lease.get("worktree") or "",
+                phase="released",
+                candidate_head=leased_head,
+                target_branch=lease.get("target_branch") or "master",
+                target_branch_sha=lease.get("target_branch_sha"),
+                last_activity=now,
+                blocker="obsolete-superseded-or-expired-lease",
+            )
+            audit_comment_body = (
+                "## Canonical obsolete reviewer lease cleanup (#691)\n\n"
+                f"- pr: #{pr_number}\n"
+                f"- lease_comment_id: {lease.get('comment_id')}\n"
+                f"- session_id: {owner_session}\n"
+                f"- leased_head: {leased_head}\n"
+                f"- current_head: {current_head}\n"
+                f"- expires_at: {lease.get('expires_at')}\n"
+                f"- classification: {classification}\n"
+                f"- terminal_review_verdict: "
+                f"{(terminal_review or {}).get('verdict')}\n"
+                f"- tool: {CLEANUP_OBSOLETE_LEASE_TOOL}\n"
+                "- action: posted terminal phase=released lease marker; "
+                "did not transfer validation, decision, or workflow proof; "
+                "did not repoint lease to the new head; did not delete history\n"
+            )
+            reasons.append(
+                f"cleanup eligible ({classification}); post terminal released "
+                "marker via sanctioned tool"
+            )
+        else:
+            reasons.extend(fail_closed_reasons)
+
+    return {
+        "pr_number": pr_number,
+        "classification": classification,
+        "blocker_kind": classification if not cleanup_allowed else "none",
+        "cleanup_allowed": cleanup_allowed,
+        "mutation_allowed": False,  # never grants review mutation
+        "mutation_eligibility": "prohibited" if not cleanup_allowed else "cleanup_only",
+        "exact_next_action": (
+            NEXT_ACTION_CLEANUP_OBSOLETE_LEASE
+            if cleanup_allowed
+            else (
+                NEXT_ACTION_WAIT
+                if classification
+                in {
+                    "foreign_active_current_head",
+                    "ambiguous_conflicting_evidence",
+                }
+                else NEXT_ACTION_OPERATOR_AUTHORIZED_CLEANUP
+            )
+        ),
+        "cleanup_tool": CLEANUP_OBSOLETE_LEASE_TOOL,
+        "required_confirmation": cleanup_confirmation_for_pr(pr_number),
+        "leased_head": leased_head,
+        "current_head": current_head,
+        "head_superseded": head_superseded,
+        "past_expiry": past_expiry,
+        "expires_at": (lease or {}).get("expires_at") if lease else None,
+        "expires_parseable": expires_parseable,
+        "terminal_review": terminal_review,
+        "terminal_review_present": terminal_review is not None,
+        "worktree_state": {
+            "path": (lease or {}).get("worktree") if lease else None,
+            "exists": worktree_exists,
+            "clean": worktree_clean,
+            "has_unpreserved_work": worktree_has_unpreserved_work,
+        },
+        "owner_session_evidence": {
+            "session_id": (lease or {}).get("session_id") if lease else None,
+            "reviewer_identity": (
+                (lease or {}).get("reviewer_identity") if lease else None
+            ),
+            "process_alive": owner_process_alive,
+            "pid_observed": owner_pid_observed,
+            "pid_is_not_ownership_proof": True,
+            "requesting_session_is_owner": bool(
+                lease
+                and req_session
+                and (lease.get("session_id") or "").strip() == req_session
+            ),
+        },
+        "active_lease": lease,
+        "release_body": release_body,
+        "audit_comment_body": audit_comment_body,
+        "controller_recovery_authorized": controller_recovery_authorized,
+        "reasons": reasons if reasons else fail_closed_reasons,
+        "fail_closed_reasons": fail_closed_reasons,
+        "forbidden": [
+            "manual comment deletion",
+            "database edits",
+            "mtime manipulation",
+            "PID-based ownership",
+            "direct session-state seeding",
+            "lease stealing or adoption",
+            "repointing old lease to new head",
+            "transfer of validation or decision state",
+            "worktree reuse by replacement reviewer",
+        ],
+    }
+
+
+def diagnose_reviewer_pr_lease_handoff(
+    comments: list[dict],
+    *,
+    pr_number: int,
+    current_session_id: str | None,
+    current_reviewer_identity: str | None,
+    proposed_worktree: str | None = None,
+    env_bound_worktree: str | None = None,
+    instructed_session_id: str | None = None,
+    instructed_comment_id: int | None = None,
+    current_head_sha: str | None = None,
+    formal_reviews: list[dict] | None = None,
+    worktree_exists: bool | None = None,
+    worktree_clean: bool | None = None,
+    owner_process_alive: bool | None = None,
+    now: datetime | None = None,
+) -> dict[str, Any]:
+    """Classify open-PR reviewer lease handoff and emit a canonical next action (#599, #691).
+
+    Read-only diagnosis. Never steals, releases, or adopts a foreign lease.
+    Fail-closed acquisition rules for active foreign leases remain intact.
+
+    Returns a structured diagnosis with:
+    - classification (including #691 superseded/expired distinctions)
+    - next_action (including cleanup_obsolete_reviewer_comment_lease)
+    - active_lease identity fields when present
+    - worktree_binding match result
+    - instructed-lease mismatch flags
+    - cleanup tool/confirmation when cleanup-eligible
+    """
+    now = now or datetime.now(timezone.utc)
+    session_id = (current_session_id or "").strip()
+    identity = (current_reviewer_identity or "").strip()
+    instructed_sid = (instructed_session_id or "").strip()
+    reasons: list[str] = []
+    current_head = _normalize_sha(current_head_sha)
+
+    active = find_active_reviewer_lease(comments, pr_number=pr_number, now=now)
+    # Also surface expired non-terminal newest marker for #691 diagnosis.
+    newest_nt = find_newest_nonterminal_lease(
+        comments, pr_number=pr_number, now=now, include_expired=True
+    )
+    lease_for_class = active or newest_nt
+    session_lease = get_session_lease()
+
+    # Worktree binding: env-bound vs proposed vs active lease worktree.
+    env_wt = _norm_path(env_bound_worktree)
+    prop_wt = _norm_path(proposed_worktree)
+    lease_wt = _norm_path((lease_for_class or {}).get("worktree") if lease_for_class else None)
+    binding_mismatch = False
+    binding_details: dict[str, Any] = {
+        "env_bound_worktree": env_bound_worktree or None,
+        "proposed_worktree": proposed_worktree or None,
+        "lease_worktree": (lease_for_class or {}).get("worktree") if lease_for_class else None,
+        "match": True,
+    }
+    paths = [p for p in (env_wt, prop_wt, lease_wt) if p]
+    if len(paths) >= 2 and len(set(paths)) > 1:
+        binding_mismatch = True
+        binding_details["match"] = False
+        reasons.append(
+            "worktree binding mismatch: env/proposed/lease worktree paths disagree"
+        )
+
+    # Instructed lease gone while a different lease is active (PR #592-style).
+    instructed_missing_with_replacement = False
+    if instructed_sid or instructed_comment_id is not None:
+        if not lease_for_class:
+            reasons.append(
+                "instructed lease is gone and no active replacement lease remains"
+            )
+        else:
+            owner = (lease_for_class.get("session_id") or "").strip()
+            cid = lease_for_class.get("comment_id")
+            sid_mismatch = bool(instructed_sid and owner and owner != instructed_sid)
+            cid_mismatch = (
+                instructed_comment_id is not None
+                and cid is not None
+                and int(cid) != int(instructed_comment_id)
+            )
+            if sid_mismatch or cid_mismatch:
+                instructed_missing_with_replacement = True
+                reasons.append(
+                    "instructed lease is gone; a different active lease replaced it "
+                    f"(active session_id={owner}, comment_id={cid})"
+                )
+
+    # Classification + next_action.
+    classification = "no_lease"
+    next_action = NEXT_ACTION_ACQUIRE
+    cleanup_hint: dict[str, Any] | None = None
+
+    if lease_for_class:
+        owner = (lease_for_class.get("session_id") or "").strip()
+        freshness = lease_for_class.get("freshness") or classify_lease_freshness(
+            lease_for_class, now=now
+        )
+        owner_identity = (lease_for_class.get("reviewer_identity") or "").strip()
+        is_own = bool(session_id and owner and owner == session_id)
+        # Same identity alone is NOT ownership for resume; session_id must match.
+        same_identity = bool(
+            identity and owner_identity and identity == owner_identity
+        )
+        leased_head = _normalize_sha(lease_for_class.get("candidate_head"))
+        head_superseded = bool(
+            current_head and leased_head and current_head != leased_head
+        )
+        head_current = bool(
+            current_head and leased_head and current_head == leased_head
+        )
+        past_expiry = bool(lease_for_class.get("expired")) or freshness == "expired"
+        if not past_expiry:
+            past_expiry = _lease_expired(lease_for_class, now=now)
+        terminal = formal_terminal_review_for_head(
+            formal_reviews, leased_head=leased_head
+        )
+
+        if is_own and freshness in {"active", "stale_warning"} and not past_expiry:
+            classification = "own_active"
+            next_action = NEXT_ACTION_RESUME_EXACT_OWNER_SESSION
+        elif is_own and (freshness in {"reclaimable", "expired"} or past_expiry):
+            classification = "own_expired"
+            next_action = NEXT_ACTION_RELEASE_EXPIRED_LEASE
+            reasons.append(
+                f"own lease freshness is '{freshness}'; release via "
+                "gitea_release_reviewer_pr_lease then re-acquire"
+            )
+        elif not is_own and head_superseded and terminal and past_expiry:
+            classification = "foreign_expired_superseded_head"
+            next_action = NEXT_ACTION_CLEANUP_OBSOLETE_LEASE
+            reasons.append(
+                "foreign lease expired and pinned to superseded head with "
+                "formal terminal review; use guarded obsolete-lease cleanup"
+            )
+        elif not is_own and head_superseded and terminal and not past_expiry:
+            classification = "foreign_completed_superseded_head"
+            next_action = NEXT_ACTION_CLEANUP_OBSOLETE_LEASE
+            reasons.append(
+                "foreign lease pinned to superseded head with completed formal "
+                "review; use guarded obsolete-lease cleanup (do not wait indefinitely)"
+            )
+        elif not is_own and head_superseded and not terminal:
+            classification = "ambiguous_conflicting_evidence"
+            next_action = NEXT_ACTION_WAIT
+            reasons.append(
+                "lease head superseded but no formal terminal review for leased "
+                "head; fail closed / wait (no indefinite steal)"
+            )
+        elif not is_own and head_current and past_expiry:
+            classification = "foreign_expired_current_head"
+            if owner_process_alive is False and worktree_clean is True:
+                classification = "orphaned_owner_missing"
+                next_action = NEXT_ACTION_CLEANUP_OBSOLETE_LEASE
+                reasons.append(
+                    "foreign expired lease on current head; owner process absent "
+                    "and worktree clean — guarded orphan cleanup eligible"
+                )
+            elif owner_process_alive is False and worktree_clean is False:
+                classification = "ambiguous_conflicting_evidence"
+                next_action = NEXT_ACTION_WAIT
+                reasons.append(
+                    "owner process absent but worktree dirty; cleanup denied"
+                )
+            else:
+                next_action = NEXT_ACTION_CLEANUP_OBSOLETE_LEASE
+                reasons.append(
+                    "foreign expired lease on current head; use guarded cleanup "
+                    "when worktree/process evidence permits"
+                )
+        elif not is_own and freshness in {"active", "stale_warning"} and not past_expiry:
+            classification = (
+                "foreign_active_current_head" if head_current or not current_head
+                else "foreign_active"
+            )
+            next_action = NEXT_ACTION_WAIT
+            reasons.append(
+                f"foreign active reviewer lease (session_id={owner}, "
+                f"phase={lease_for_class.get('phase')}, freshness={freshness}); "
+                "do not submit; do not steal"
+            )
+            if same_identity:
+                reasons.append(
+                    "lease identity matches current reviewer but session_id differs; "
+                    "resume only from the exact owner session_id or wait"
+                )
+        elif not is_own and freshness == "reclaimable":
+            classification = "foreign_reclaimable"
+            next_action = NEXT_ACTION_RELEASE_EXPIRED_LEASE
+            reasons.append(
+                f"foreign reclaimable lease (session_id={owner}); clear only via "
+                "sanctioned gitea_release_reviewer_pr_lease when reclaimable"
+            )
+        elif not is_own and (freshness == "expired" or past_expiry):
+            classification = "foreign_expired"
+            next_action = NEXT_ACTION_RELEASE_EXPIRED_LEASE
+            reasons.append(
+                f"foreign expired lease (session_id={owner}); use sanctioned release "
+                f"or {CLEANUP_OBSOLETE_LEASE_TOOL}"
+            )
+        else:
+            classification = "foreign_active"
+            next_action = NEXT_ACTION_WAIT
+            reasons.append(
+                f"unclassified active lease state (session_id={owner}, "
+                f"freshness={freshness}); wait fail-closed"
+            )
+
+        if instructed_missing_with_replacement and classification.startswith(
+            "foreign"
+        ):
+            # Preserve #691 cleanup path when superseded/expired; otherwise
+            # keep the PR #592 instructed-missing label.
+            if next_action != NEXT_ACTION_CLEANUP_OBSOLETE_LEASE:
+                classification = "instructed_lease_missing_with_replacement"
+                if next_action == NEXT_ACTION_WAIT:
+                    reasons.append(
+                        "replacement foreign lease is active — wait; "
+                        "operator_authorized_cleanup only with explicit operator authority"
+                    )
+
+        if next_action == NEXT_ACTION_CLEANUP_OBSOLETE_LEASE:
+            cleanup_hint = {
+                "cleanup_tool": CLEANUP_OBSOLETE_LEASE_TOOL,
+                "required_confirmation": cleanup_confirmation_for_pr(pr_number),
+                "controller_recovery_authorized_required": True,
+                "leased_head": leased_head,
+                "current_head": current_head,
+                "expires_at": lease_for_class.get("expires_at"),
+                "terminal_review_verdict": (terminal or {}).get("verdict"),
+            }
+    else:
+        classification = "no_lease"
+        next_action = NEXT_ACTION_ACQUIRE
+        reasons.append("no active reviewer lease; acquire via gitea_acquire_reviewer_pr_lease")
+
+    # Binding mismatch is a first-class blocker before submit, but does not
+    # erase foreign-lease wait/release guidance. Override next_action only when
+    # the session would otherwise be free to acquire or resume (submit path).
+    if binding_mismatch:
+        if next_action in {
+            NEXT_ACTION_ACQUIRE,
+            NEXT_ACTION_RESUME_EXACT_OWNER_SESSION,
+        }:
+            classification = "worktree_binding_mismatch"
+            next_action = NEXT_ACTION_REPAIR_WORKTREE_BINDING
+        else:
+            reasons.append(
+                "also repair worktree binding before submit "
+                f"(next_action remains {next_action})"
+            )
+
+    lease_summary = None
+    if lease_for_class:
+        lease_summary = {
+            "comment_id": lease_for_class.get("comment_id"),
+            "session_id": lease_for_class.get("session_id"),
+            "phase": lease_for_class.get("phase"),
+            "candidate_head": lease_for_class.get("candidate_head"),
+            "expires_at": lease_for_class.get("expires_at"),
+            "last_activity": lease_for_class.get("last_activity"),
+            "freshness": lease_for_class.get("freshness")
+            or classify_lease_freshness(lease_for_class, now=now),
+            "reviewer_identity": lease_for_class.get("reviewer_identity"),
+            "profile": lease_for_class.get("profile"),
+            "worktree": lease_for_class.get("worktree"),
+            "blocker": lease_for_class.get("blocker"),
+        }
+
+    return {
+        "pr_number": pr_number,
+        "classification": classification,
+        "blocker_kind": classification,
+        "next_action": next_action,
+        "exact_next_action": next_action,
+        "active_lease": lease_summary,
+        "session_lease": session_lease,
+        "worktree_binding": binding_details,
+        "leased_head": (lease_summary or {}).get("candidate_head"),
+        "current_head": current_head,
+        "expires_at": (lease_summary or {}).get("expires_at"),
+        "terminal_review_state": (
+            formal_terminal_review_for_head(
+                formal_reviews,
+                leased_head=(lease_summary or {}).get("candidate_head"),
+            )
+            if lease_summary
+            else None
+        ),
+        "worktree_state": {
+            "exists": worktree_exists,
+            "clean": worktree_clean,
+            "path": (lease_summary or {}).get("worktree"),
+        },
+        "owner_session_evidence": {
+            "session_id": (lease_summary or {}).get("session_id"),
+            "process_alive": owner_process_alive,
+            "pid_is_not_ownership_proof": True,
+        },
+        "cleanup": cleanup_hint,
+        "cleanup_tool": (cleanup_hint or {}).get("cleanup_tool"),
+        "required_confirmation": (cleanup_hint or {}).get("required_confirmation"),
+        "instructed_session_id": instructed_session_id,
+        "instructed_comment_id": instructed_comment_id,
+        "instructed_lease_missing_with_replacement": instructed_missing_with_replacement,
+        "mutation_allowed": (
+            next_action == NEXT_ACTION_RESUME_EXACT_OWNER_SESSION
+            and not binding_mismatch
+            and bool(session_lease)
+        ),
+        "mutation_eligibility": (
+            "allowed"
+            if (
+                next_action == NEXT_ACTION_RESUME_EXACT_OWNER_SESSION
+                and not binding_mismatch
+                and bool(session_lease)
+            )
+            else (
+                "cleanup_only"
+                if next_action == NEXT_ACTION_CLEANUP_OBSOLETE_LEASE
+                else "prohibited"
+            )
+        ),
+        "reasons": reasons,
+        "forbidden": [
+            "manual lock deletion",
+            "raw API bypass",
+            "mtime manipulation",
+            "direct _SESSION_LEASE seeding",
+            "silent foreign lease steal",
+            "PID-based ownership",
+            "repointing old lease to new head",
+            "transfer of validation or decision state",
+        ],
+    }

reviewer_worktree.py

@@ -27,6 +27,20 @@ _READONLY_REVIEWER_GIT = re.compile(
 _GIT_INVOCATION = re.compile(r"\bgit\b", re.IGNORECASE)
 
 
+# #673: Canonical pattern for identifying review worktrees under branches/.
+REVIEW_WORKTREE_RE = re.compile(
+    r"branches/(?:review-pr\d+[\w/-]*|merge-simulation-pr\d+|review-[\w-]+)",
+    re.IGNORECASE,
+)
+
+
+def is_review_worktree_path(path: str) -> bool:
+    """True when the path belongs to a reviewer or simulation worktree."""
+    normalized = (path or "").replace("\\", "/")
+    return bool(REVIEW_WORKTREE_RE.search(normalized))
+
+
+
 def parse_dirty_tracked_files(porcelain: str) -> list[str]:
     """Return tracked paths with local modifications from ``git status --porcelain``.
 

reviewer_worktree_ownership.py

@@ -5,10 +5,9 @@ from __future__ import annotations
 import re
 from typing import Any
 
-_SESSION_OWNED_RE = re.compile(
-    r"branches/(?:review-pr\d+[\w/-]*|review-[\w-]+)",
-    re.IGNORECASE,
-)
+from reviewer_worktree import REVIEW_WORKTREE_RE
+
+_SESSION_OWNED_RE = REVIEW_WORKTREE_RE
 _WORKTREE_PATH_RE = re.compile(
     r"(?:review worktree path|worktree path|session-owned worktree)\s*:\s*(\S+)",
     re.IGNORECASE,

role_session_router.py

@@ -55,7 +55,6 @@ def skip_python_scan_walk_root(project_root: str, walk_root: str) -> bool:
 
 REVIEWER_TASKS = frozenset({
     "review_pr",
-    "merge_pr",
     "blind_pr_queue_review",
     "pr_queue_cleanup",
     "pr-queue-cleanup",
@@ -63,6 +62,10 @@ REVIEWER_TASKS = frozenset({
     "approve_pr",
 })
 
+MERGER_TASKS = frozenset({
+    "merge_pr",
+})
+
 AUTHOR_TASKS = frozenset({
     "create_issue",
     "comment_issue",
@@ -80,6 +83,7 @@ AUTHOR_TASKS = frozenset({
 })
 
 RECONCILER_TASKS = frozenset({
+    "cleanup_merged_pr_branch",
     "reconcile_already_landed_pr",
     "reconcile_already_landed",
     "reconcile-landed-pr",
@@ -97,7 +101,7 @@ TASK_REQUIRED_ROLE = {
     "address_pr_change_requests": "author",
     "delete_branch": "author",
     "review_pr": "reviewer",
-    "merge_pr": "reviewer",
+    "merge_pr": "merger",
     "blind_pr_queue_review": "reviewer",
     "pr_queue_cleanup": "reviewer",
     "pr-queue-cleanup": "reviewer",
@@ -109,9 +113,13 @@ TASK_REQUIRED_ROLE = {
     "reconcile_already_landed_pr": "reconciler",
     "reconcile_already_landed": "reconciler",
     "reconcile-landed-pr": "reconciler",
+    "cleanup_merged_pr_branch": "reconciler",
     # #309: reconciler tasks close already-landed PRs/issues only.
     "reconcile_close_landed_pr": "reconciler",
     "reconcile_close_landed_issue": "reconciler",
+    "reconcile_close_superseded_pr": "reconciler",
+    "reconcile_close_satisfied_issue": "reconciler",
+    "reconcile_create_followup_issue": "reconciler",
 }
 
 WRONG_ROLE_REVIEWER_MSG = (
@@ -123,6 +131,10 @@ WRONG_ROLE_RECONCILER_MSG = (
     "MCP namespace/profile with exact close capability."
 )
 
+WRONG_ROLE_MERGER_MSG = (
+    "Wrong role/session for merger task. Launch merger MCP namespace."
+)
+
 _session_last_route: dict | None = None
 
 
@@ -238,6 +250,25 @@ def route_task_session(
         _record_route(result)
         return result
 
+    if required_role == "merger":
+        result = {
+            "task_type": task_type,
+            "required_role": required_role,
+            "active_role": active_role_kind,
+            "active_profile": active_profile,
+            "route_result": ROUTE_WRONG_ROLE,
+            "downstream_allowed": False,
+            "reasons": [
+                WRONG_ROLE_MERGER_MSG,
+                "Merger tasks cannot run in author or reviewer sessions.",
+            ],
+            "message": WRONG_ROLE_MERGER_MSG,
+            "runtime_switching_supported": runtime_switching_supported,
+            "profile_switch_blocked": not runtime_switching_supported,
+        }
+        _record_route(result)
+        return result
+
     if required_role == "author":
         route = ROUTE_TO_AUTHOR
         message = (
@@ -406,4 +437,4 @@ def assess_infra_stop(project_root: str | None = None) -> dict:
 
 def check_mid_merge(project_root: str | None = None) -> bool:
     """Return True if the repository is mid-merge, mid-rebase, or has conflict markers."""
-    return assess_infra_stop(project_root)["infra_stop"]
+    return assess_infra_stop(project_root)["infra_stop"]

scripts/ci-webui-check

@@ -0,0 +1,53 @@
+#!/usr/bin/env bash
+# Path-filtered CI gate for the internal web UI (#436).
+# Runs the hermetic web UI unittest suite when a change touches UI surfaces.
+set -euo pipefail
+
+script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+repo_root="$(cd "$script_dir/.." && pwd)"
+cd "$repo_root"
+
+if [[ "${WEBUI_CI_FORCE:-}" == "1" ]]; then
+  exec "$script_dir/test-webui"
+fi
+
+base_ref="${WEBUI_CI_BASE_REF:-}"
+if [[ -z "$base_ref" ]]; then
+  if git rev-parse --verify prgs/master >/dev/null 2>&1; then
+    base_ref="$(git merge-base HEAD prgs/master 2>/dev/null || true)"
+  fi
+  if [[ -z "$base_ref" ]]; then
+    base_ref="${GITHUB_BASE_REF:-${CHANGE_TARGET:-master}}"
+    if git rev-parse --verify "origin/$base_ref" >/dev/null 2>&1; then
+      base_ref="$(git merge-base HEAD "origin/$base_ref")"
+    elif git rev-parse --verify "$base_ref" >/dev/null 2>&1; then
+      base_ref="$(git merge-base HEAD "$base_ref")"
+    else
+      base_ref="HEAD~1"
+    fi
+  fi
+fi
+
+changed="$(git diff --name-only "$base_ref" HEAD 2>/dev/null || true)"
+if [[ -z "$changed" ]]; then
+  echo "ci-webui-check: no changed files vs $base_ref; skipping web UI suite"
+  exit 0
+fi
+
+python_bin="${WEBUI_TEST_PYTHON:-python3}"
+if [[ -x "$repo_root/venv/bin/python" ]]; then
+  python_bin="$repo_root/venv/bin/python"
+fi
+
+if printf '%s\n' "$changed" | "$python_bin" -c "
+import sys
+from webui.ci_paths import should_run_webui_ci
+paths = [line.strip() for line in sys.stdin if line.strip()]
+raise SystemExit(0 if should_run_webui_ci(paths) else 1)
+"; then
+  echo "ci-webui-check: web UI surface changed; running suite"
+  exec "$script_dir/test-webui"
+fi
+
+echo "ci-webui-check: no web UI paths in diff; skipping suite"
+exit 0

scripts/install-codex-workflow-skill.sh

@@ -0,0 +1,82 @@
+#!/usr/bin/env bash
+# Install canonical Gitea workflow skill into Codex skills dir (#551).
+# Symlinks the in-repo skills/llm-project-workflow package under the names
+# gitea-workflow, llm-project-workflow, and git-pr-workflows.
+set -euo pipefail
+
+usage() {
+  cat <<'EOF'
+usage: scripts/install-codex-workflow-skill.sh [--dry-run] [--skills-dir DIR]
+
+Symlink the portable skills/llm-project-workflow package into the Codex
+skills directory under the canonical names:
+  gitea-workflow
+  llm-project-workflow
+  git-pr-workflows
+
+Defaults:
+  skills dir: $CODEX_HOME/skills or ~/.codex/skills
+  source:     <repo>/skills/llm-project-workflow
+EOF
+}
+
+dry_run=0
+skills_dir=""
+while [[ "${1:-}" == --* ]]; do
+  case "$1" in
+    --dry-run) dry_run=1 ;;
+    --skills-dir)
+      shift
+      skills_dir="${1:-}"
+      ;;
+    --help|-h) usage; exit 0 ;;
+    *) usage >&2; exit 2 ;;
+  esac
+  shift
+done
+
+script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+repo_root="$(cd "$script_dir/.." && pwd)"
+source_skill="$repo_root/skills/llm-project-workflow"
+
+if [[ ! -f "$source_skill/SKILL.md" ]]; then
+  echo "Error: missing $source_skill/SKILL.md" >&2
+  exit 1
+fi
+
+if [[ -z "$skills_dir" ]]; then
+  if [[ -n "${CODEX_HOME:-}" ]]; then
+    skills_dir="$CODEX_HOME/skills"
+  else
+    skills_dir="${HOME}/.codex/skills"
+  fi
+fi
+
+names=(gitea-workflow llm-project-workflow git-pr-workflows)
+
+echo "source: $source_skill"
+echo "codex skills dir: $skills_dir"
+
+if [[ "$dry_run" -eq 0 ]]; then
+  mkdir -p "$skills_dir"
+fi
+
+for name in "${names[@]}"; do
+  target="$skills_dir/$name"
+  if [[ "$dry_run" -eq 1 ]]; then
+    echo "dry-run: ln -sfn $source_skill $target"
+    continue
+  fi
+  if [[ -e "$target" || -L "$target" ]]; then
+    if [[ -L "$target" ]]; then
+      rm -f "$target"
+    else
+      echo "Error: $target exists and is not a symlink (refuse to overwrite)" >&2
+      exit 1
+    fi
+  fi
+  ln -sfn "$source_skill" "$target"
+  echo "linked $target -> $source_skill"
+done
+
+echo "done. Restart Codex so skills reload."

scripts/test-webui

@@ -0,0 +1,14 @@
+#!/usr/bin/env bash
+set -euo pipefail
+script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+repo_root="$(cd "$script_dir/.." && pwd)"
+cd "$repo_root"
+
+python_bin="${WEBUI_TEST_PYTHON:-python3}"
+if [[ -x "$repo_root/venv/bin/python" ]]; then
+  python_bin="$repo_root/venv/bin/python"
+fi
+
+pattern="${WEBUI_TEST_PATTERN:-test_webui_*.py}"
+export WEBUI_TEST_OFFLINE="${WEBUI_TEST_OFFLINE:-1}"
+exec "$python_bin" -m unittest discover -s tests -p "$pattern" "$@"

skills/gitea-workflow/SKILL.md

@@ -0,0 +1,37 @@
+---
+name: gitea-workflow
+description: >-
+  Canonical alias for the Gitea/LLM project workflow router. Same skill as
+  llm-project-workflow. Use at the start of any Gitea-Tools author, review,
+  merge, reconcile, or issue-filing task. Trigger terms: gitea-workflow,
+  llm-project-workflow, git-pr-workflows, gitea, PR review, work-issue.
+---
+
+# gitea-workflow (canonical alias)
+
+This directory is the **stable name** for controller prompts and Codex mounts
+(`gitea-workflow`). The portable implementation lives at:
+
+**[`../llm-project-workflow/SKILL.md`](../llm-project-workflow/SKILL.md)**
+
+## Required action
+
+1. Open and follow `skills/llm-project-workflow/SKILL.md` (router).
+2. Load the matching workflow under `skills/llm-project-workflow/workflows/`.
+3. Do not mutate git/Gitea until the workflow is loaded.
+
+## Multi-runtime names (must resolve identically)
+
+| Name | Role |
+|------|------|
+| `gitea-workflow` | Canonical controller / Codex skill name |
+| `llm-project-workflow` | Portable in-repo skill package |
+| `git-pr-workflows` | Legacy alias |
+
+Install for Codex:
+
+```bash
+./scripts/install-codex-workflow-skill.sh
+```
+
+Preflight via MCP: `mcp_check_workflow_skill_preflight`.

skills/llm-project-workflow/SKILL.md

@@ -6,6 +6,8 @@ description: >-
   report schema. Use at the start of any implementation, review, merge,
   reconciliation, or issue-filing task.
 ---
+> **Also known as:** `gitea-workflow`, `git-pr-workflows` (canonical multi-runtime names — see docs/workflow-skill-mount.md / #551).
+
 
 # LLM Project Workflow Skill
 
@@ -30,13 +32,44 @@ workflow file.
   mutation.
 - A nearby capability does not count.
 - Do not self-review or self-merge.
+- **Never push a stable branch directly.** Worker sessions (author/reviewer/merger)
+  must never run `git push <remote> master` — or `main`/`dev`/other stable refs,
+  including refspecs (`HEAD:master`), `--force`, `--delete`, or `--dry-run` no-op
+  probes — and must never commit on the root/control checkout outside an issue
+  feature branch. Stable-branch updates land ONLY through sanctioned Gitea merge
+  tooling (`gitea_merge_pr`) or an explicitly authorized reconciler path. A
+  detected attempt marks the session workflow-contaminated (#671) and fails
+  closed on review/merge/close/completion until a reconciler audits and clears
+  it. `git fetch` / `git pull --ff-only` and feature-branch pushes stay allowed.
 - Do not mix modes in one run.
+- **BLOCKED + DIAGNOSE default rule (required):** If any required workflow step, skill, tool, capability, preflight, instruction, profile, worktree binding, or terminal/MCP operation cannot be performed or loaded (including the canonical ones listed in this skill and its loaded workflow), immediately enter `BLOCKED + DIAGNOSE`. Stop before any git or Gitea mutation. Diagnose using the standard template in [`templates/blocked-diagnose-report.md`](templates/blocked-diagnose-report.md). Attempt *only* safe non-mutating recovery. Report using the template. Do not continue, use fallbacks, or treat the missing requirement as harmless.
 - If the required workflow cannot be loaded, stop and produce a recovery handoff
-  only.
+  only (see BLOCKED + DIAGNOSE rule above).
 - Final report must use the schema for the loaded workflow.
 - If a task requires a different mode, stop and produce a handoff for the
   correct workflow.
 
+## Covered blocker classes (BLOCKED + DIAGNOSE must trigger for these)
+
+- missing required skill or workflow guide (e.g. gitea-workflow, llm-project-workflow)
+- broken terminal/tool runner or shell spawn failure
+- MCP capability failure, reset, or deadlock (e.g. preflight state cleared)
+- wrong profile or role for the requested operation
+- dirty or misbound worktree (root checkout or non-branches/ path)
+- root checkout mutation risk
+- stable-branch push contamination (#671): a direct `git push <remote> master`
+  (or `main`/`dev`) equivalent, or a root/control-checkout commit not on an issue
+  feature branch, marks the session workflow-contaminated; review/merge/close/
+  completion mutations then fail closed until a reconciler audits and clears it
+  (`gitea_audit_stable_branch_contamination action=clear`, reconciler-only)
+- mutation guard failure (e.g. branches-only guard)
+- missing required MCP tool/schema or operation
+- stale or inconsistent runtime state (e.g. lease vs actual, dirty state disagreement)
+- unavailable project instructions or checked-in guides
+- any other failure of a step the current workflow or controller prompt declares "required"
+
+**Prohibited unless controller authorizes in writing for this instance:** temp scripts, direct API fallback, MCP internals, direct imports, in-memory state restoration, manual bypasses, or any continuation that hides the blocker. All such cases must be reported as process/tooling defects.
+
 ## Mode isolation
 
 A run that starts in `review-merge-pr` mode may not create process issues,
@@ -99,6 +132,42 @@ The main project checkout is a stable control checkout on `master`, `main`, or
 If `cwd` is not inside `branches/`, stop before any file edit, test write,
 commit, merge, rebase, or cleanup. The main checkout is orchestration-only.
 
+## Stable Branch Push Protection (#671)
+
+Worker sessions must **never** publish a stable branch directly. This is the
+prevention hardening for the #670 incident (a bare direct-to-master commit
+`2fa97c26` and a PR #654 merger `git push prgs master` attempt).
+
+**Forbidden for author/reviewer/merger sessions:**
+
+- `git push <remote> master` (and `main`, `dev`, `develop`, `development`),
+  including refspecs (`HEAD:master`, `+refs/heads/x:refs/heads/master`),
+  `--force`, `--delete` / `:master`, and `--dry-run`/`-n` no-op probes (a
+  dry-run still proves intent and contaminates the session).
+- Local commits on the root/control checkout that are not carried by an issue
+  feature branch under `branches/`.
+
+**Allowed (never blocked):**
+
+- `git fetch` and `git pull --ff-only` of master into the control checkout when
+  authorized for sync.
+- Feature-branch pushes to non-stable refs (`git push <remote> fix/issue-N-...`).
+- Sanctioned merges via `gitea_merge_pr` / the Gitea API merge endpoint — the
+  **only** way stable branches advance.
+
+**What happens on a detected attempt:** the session is marked
+workflow-contaminated (durable `stable_branch_contamination` marker, redacted
+command summary + session id + remote + ref). While contaminated, all
+review / merge / close / issue-completion mutations fail closed. `comment_issue`
+and `lock_issue` remain allowed so the contaminated worker can post the durable
+audit comment and hand off. Contamination **cannot be self-cleared** — only a
+reconciler audit (`gitea_audit_stable_branch_contamination action=clear`) may
+clear it.
+
+Tooling: call `gitea_record_stable_branch_push_attempt` to classify/record a
+proposed push before running it; `gitea_audit_stable_branch_contamination` to
+inspect or (reconciler-only) clear the marker.
+
 ## Shell Spawn Hard-Stop Rule
 
 `exit_code: -1` with empty stdout/stderr means the shell failed to spawn — not a
@@ -167,6 +236,7 @@ Ready-to-copy task prompts live in [`templates/`](templates/):
 - [`reconcile-closed-not-merged-pr.md`](templates/reconcile-closed-not-merged-pr.md)
 - [`worktree-cleanup.md`](templates/worktree-cleanup.md)
 - [`release-tag.md`](templates/release-tag.md)
+- [`canonical-state-comments.md`](templates/canonical-state-comments.md)
 
 ## Adapting to a project
 
@@ -184,6 +254,18 @@ Releases follow SemVer from remote `master` only, after full test suite passes.
 See [`templates/release-tag.md`](templates/release-tag.md) and
 `scripts/release-tag`.
 
+## Proof: missing required workflow steps stop before mutation
+
+- The llm-project-workflow router (this file) and every loaded workflow (work-issue.md, review-merge-pr.md, create-issue.md, etc.) now declare at the top: if required step/skill/tool/capability/instruction/profile/worktree binding/preflight fails, STOP, state BLOCKED, use blocked-diagnose-report.md template, only non-mutating recovery.
+- Controller prompts (start-issue.md, review-pr.md, merge-pr.md, recover-bad-state.md, etc.) and the runbooks (docs/llm-workflow-runbooks.md) explicitly require the same and prohibit unsafe fallbacks.
+- MCP guards (branches-only mutation guard #274, worktree binding #510, preflight purity, role checks, lease gates, gitea_lock_issue, etc.) plus the "prove before mutation" rules ensure that a BLOCKED state prevents git/Gitea mutations.
+- When a skill/guide/tool is missing (e.g. gitea-workflow not mounted for a runtime), the load step in the router/prompt fails the "required" check → BLOCKED + report before any gitea_* call or git command that mutates.
+- Terminal/shell failures, capability deadlocks, wrong profile, dirty/misbound worktree, root risk, guard failures, missing schema, stale state, unavailable instructions all map to the covered blocker classes and trigger the same stop + report.
+- No code path in the canonical workflows allows continuation past a declared required step without the BLOCKED report.
+- See also: Global LLM Worktree Rule, Shell Spawn Hard-Stop Rule, Identity and profile safety, Subagent Tool-Budget Guardrails, and the explicit prohibition list in Universal rules.
+
+Tests / proof docs updated in this change + runbooks. Full relevant test runs (see PR handoff) pass; `git diff --check` clean. Missing-step cases are now documented to fail closed before mutation.
+
 ## Bootstrap Review Path (#557)
 
 Self-hosted MCP workflow fixes can deadlock live review daemons. Do not bypass
@@ -195,4 +277,3 @@ If and only if a controller posts a durable `BOOTSTRAP REVIEW AUTHORIZATION
 `docs/bootstrap-review-path.md`
 
 Otherwise stop with BLOCKED + DIAGNOSE. Bootstrap never weakens normal PR gates.
-

skills/llm-project-workflow/templates/blocked-diagnose-report.md

@@ -0,0 +1,52 @@
+# Blocker Report Template (BLOCKED + DIAGNOSE)
+
+Use this exact structure whenever a required workflow step cannot be performed. Emit this report and stop. Do not proceed to mutation or fallback unless a controller explicitly authorizes an exception in writing.
+
+## Required step
+<Describe the exact step, skill, tool, capability, instruction, profile, or preflight that was required. Include the canonical name and where it is defined (e.g. gitea-workflow skill, specific workflow file, gitea_xxx tool).>
+
+## Observed failure
+<Exact symptom, error message, missing output, guard error, 404, schema error, dirty state, wrong profile, etc. Quote relevant output or tool response.>
+
+## Expected behavior
+<What the workflow/docs/prompts say should happen. Reference the specific rule, template, or preflight that requires this step.>
+
+## Checks performed
+- List every verification attempted (e.g. gitea_whoami, resolve_task_capability, ls skills/, git status, mcp_list_*, worktree list, etc.)
+- Note any discrepancies found (e.g. skill not mounted for this runtime, capability not in profile, cwd not under branches/, etc.)
+
+## Safe recovery attempted
+- Only non-mutating actions (reads, lists, views, status, whoami, resolve, fetch --dry, etc.)
+- List what was tried and the result.
+- If no safe recovery possible, state that explicitly.
+
+## Likely classification
+Choose one or more:
+- missing required skill or workflow guide
+- broken terminal/tool runner
+- MCP capability failure or deadlock
+- wrong profile or role
+- dirty or misbound worktree
+- root checkout mutation risk
+- mutation guard failure
+- missing required MCP tool/schema
+- stale or inconsistent runtime state
+- unavailable project instructions
+- other: <describe>
+
+## Durable fix recommendation
+<Specific, actionable recommendation that fixes the root process/tooling issue (e.g. "Mount gitea-workflow skill for Codex under canonical name in ~/.codex/skills/", "Add preflight in llm-project-workflow/SKILL.md that hard-stops before any gitea_ call if X is unavailable", "Update controller prompt to require BLOCKED + this report before any fallback", "Grant capability in profile config", etc.). Do not suggest temp workarounds.>
+
+## Mutation occurred?
+- No (preferred and required unless explicitly authorized)
+- Yes — describe exactly what was mutated and why it was unavoidable after diagnosis. (This should be rare and will trigger additional review.)
+
+## Single next action
+<One concrete next step for the current actor (e.g. "Controller to approve or reject recovery", "File follow-up issue #XXX for skill mounting", "Re-launch session from clean branches/ worktree after skill installed", "Stop and wait for profile update").>
+
+---
+
+**Rule reminder (do not bypass):**
+If the required step is unavailable, you are BLOCKED. Diagnose using this template. Report. Stop. Unsafe fallbacks (temp scripts, direct API, MCP internals, direct imports, in-memory restoration, manual bypasses) are prohibited unless a controller has authorized them for this specific instance in a prior handoff.
+
+This report must appear in the final output / handoff before any further action.

skills/llm-project-workflow/templates/blocked-review-handoff.md

@@ -0,0 +1,22 @@
+# Blocked review submission handoff
+
+Use when `gitea_submit_pr_review` or the terminal mutation gate blocks review
+submission.
+
+```text
+## Blocked Review Submission Handoff
+
+- Tool called: gitea_submit_pr_review
+- Mutation attempted: yes
+- Mutation rejected: yes
+- No server-side state changed: confirmed
+- Proof/source: <paste exact tool error or gate reason>
+- Prior terminal mutation that consumed budget: <exact prior mutation or none>
+- Review decision (local intent): <approve | request_changes | comment only>
+- Server-side final decision marked: <yes with tool proof | no>
+- Gitea review submitted: no
+- Gitea review blocked because: <exact gate/error>
+- Review mutations: none (submission blocked)
+- MCP/Gitea mutations: <only mutations that actually occurred>
+- Safe next action: rewrite handoff with consistent ledger before posting
+```

skills/llm-project-workflow/templates/canonical-state-comments.md

@@ -0,0 +1,148 @@
+# Template: canonical state comments
+
+Use these only for workflow-changing comments. Casual discussion does not need
+the full template.
+
+## Issue
+
+```text
+## Canonical Issue State
+
+STATE:
+<ready-for-author | in-progress | blocked | PR-open | needs-review | ready-to-merge | merged | closed | superseded>
+
+WHO_IS_NEXT:
+<controller | author | reviewer | merger | reconciler | user>
+
+NEXT_ACTION:
+<specific one-sentence action>
+
+NEXT_PROMPT:
+<paste-ready prompt for the next role>
+
+WHAT_HAPPENED:
+<latest meaningful event>
+
+WHY:
+<decision rationale>
+
+RELATED_DISCUSSION:
+<link/reference or none>
+
+RELATED_PRS:
+- #...
+
+BRANCH:
+<branch or none>
+
+HEAD_SHA:
+<40-character SHA or none>
+
+VALIDATION:
+<tests/proofs or none>
+
+BLOCKERS:
+<blocker and unblock condition, or none>
+
+LAST_UPDATED_BY:
+<identity/profile/date>
+```
+
+## PR
+
+```text
+## Canonical PR State
+
+STATE:
+<needs-review | changes-requested | approved | stale-approval | ready-to-merge | merged | blocked | superseded>
+
+WHO_IS_NEXT:
+<controller | author | reviewer | merger | reconciler | user>
+
+NEXT_ACTION:
+<specific one-sentence action>
+
+NEXT_PROMPT:
+<paste-ready prompt for the next role>
+
+WHAT_HAPPENED:
+<latest meaningful event>
+
+WHY:
+<decision rationale>
+
+ISSUE:
+#...
+
+BASE:
+<branch>
+
+HEAD:
+<branch>
+
+HEAD_SHA:
+<40-character SHA>
+
+REVIEW_STATUS:
+<none | approved | changes-requested | stale | contaminated>
+
+VALIDATION:
+<tests/proofs>
+
+BLOCKERS:
+<blockers or none>
+
+SUPERSEDES:
+<PRs or none>
+
+SUPERSEDED_BY:
+<PR or none>
+
+MERGE_READY:
+<yes/no and why>
+
+LAST_UPDATED_BY:
+<identity/profile/date>
+```
+
+## Discussion
+
+```text
+## Canonical Discussion Summary
+
+STATE:
+<needs-more-discussion | ready-for-issues | issues-created | closed>
+
+WHO_IS_NEXT:
+<controller | author | reviewer | user>
+
+DECISION:
+<what was decided>
+
+WHY:
+<reasoning and tradeoffs>
+
+SUBSTANTIVE_COMMENTS:
+<count and summary>
+
+ISSUES_TO_CREATE_OR_CREATED:
+- #...
+
+DEPENDENCY_ORDER:
+<order or none>
+
+NON_GOALS:
+<non-goals>
+
+OPEN_QUESTIONS:
+<questions or none>
+
+NEXT_ACTION:
+<specific one-sentence action>
+
+NEXT_PROMPT:
+<paste-ready prompt for the next role>
+
+LAST_UPDATED_BY:
+<identity/profile/date>
+```

skills/llm-project-workflow/templates/canonical-thread-handoff.md

@@ -0,0 +1,36 @@
+# Template: Canonical Thread Handoff (CTH)
+
+Copy, fill the fields, and post as an issue or PR comment.
+
+```text
+## CTH: <Type>
+
+Status: <current workflow state>
+Next owner: <author|reviewer|merger|controller|operator>
+Current blocker: <none or exact blocker>
+Decision: <what was decided this session>
+Proof: <commands, SHAs, gate outputs, or explicit pending proof>
+Next action: <one concrete step for the next actor>
+Ready-to-paste prompt: <full prompt the next session should paste>
+```
+
+Allowed `<Type>` values:
+
+- State Handoff
+- Controller Decision
+- Author Handoff
+- Reviewer Handoff
+- Merger Handoff
+- Supersession Notice
+- Blocker
+
+Rules:
+
+- Find the latest CTH comment in the thread before starting work.
+- Treat the latest valid CTH as the current source of truth.
+- Post a new CTH when you finish, block, skip, supersede, approve, request
+  changes, or hand off.
+- A CTH summarizes workflow state; formal Gitea review verdicts remain
+  authoritative for merge gates.
+
+Full protocol: `docs/canonical-thread-handoff.md`

skills/llm-project-workflow/templates/merge-pr.md

@@ -1,4 +1,4 @@
-# Template: merge a PR (eligible reviewer only)
+# Template: merge a PR (eligible merger only)
 
 Copy, fill the `<...>` fields, and paste as the task prompt.
 
@@ -10,12 +10,17 @@ Load the canonical workflow first:
 Final report schema: `schemas/review-merge-final-report.md`.
 
 Rules (llm-project-workflow):
+- **BLOCKED + DIAGNOSE default (required):** If any required step (load workflow, lease, profile/role, capability, worktree under branches/, preflight, tool, instruction, etc.) cannot be performed, STOP. State BLOCKED. Use [`blocked-diagnose-report.md`](../templates/blocked-diagnose-report.md) template exactly. Only safe non-mutating recovery. Report. Do not continue or fallback.
+- Find the latest CTH comment on the PR/issue thread before starting work.
+  Post a new CTH: Merger Handoff (or CTH: Blocker) at session end.
+  Template: skills/llm-project-workflow/templates/canonical-thread-handoff.md
 - Repository targeting (#530): pass explicit `remote=`, `org=`, and `repo=` on
   every gitea-tools call (e.g. `remote=prgs org=Scaled-Tech-Consulting
   repo=Gitea-Tools`). A bare `remote=prgs` can resolve to the wrong default repo
   and is blocked when it disagrees with the local git remote URL.
-- Only an eligible, NON-author reviewer merges. If authenticated user == PR
+- Only an eligible, NON-author merger merges. If authenticated user == PR
   author → STOP.
+- Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
 - Do not merge unless the PR is open, mergeable, and its checks/review pass.
 - No force-merge, no bypassing branch protections.
 - If the PR is closed but `merged=false`, STOP and run reconciliation. Do not clean up.
@@ -63,10 +68,12 @@ Then run the cleanup template (worktree-cleanup.md) in a reconciler session:
 - fetch/prune; confirm main checkout is clean and current (0 0).
 
 Handoff: end with a section titled exactly `Controller Handoff` per SKILL.md
-§K (long form — a merge is always high-risk), including the review/merge role
-fields (Selected PR, Reviewer eligibility, Pinned reviewed head, Review
+§K (long form — a merge is always high-risk), including the merger role
+fields (Selected PR, Merger eligibility, Pinned reviewed head, Review
 decision, Merge result, Linked issue status, Cleanup status) plus: merge
 commit, PR metadata state/merged flag/hash, remote master hash, and the
 post-merge verification method used & verification results. Reports missing
 the handoff are downgraded (review_proofs.assess_controller_handoff).
+
+Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
 ```

skills/llm-project-workflow/templates/recover-bad-state.md

@@ -3,12 +3,15 @@
 Copy, fill the `<...>` fields, paste as the task prompt. Recovery is read-then-
 act: gather facts first, never discard unmerged work.
 
+**BLOCKED + DIAGNOSE rule (llm-project-workflow):** If at any point a required step (including state recovery itself) cannot be performed, stop immediately, use the standard [`blocked-diagnose-report.md`](blocked-diagnose-report.md) template, attempt only safe non-mutating recovery, and report. Do not continue or fallback.
+
 ```text
 Task: recover repo state for <situation>. Do not lose unmerged work.
 
 Rules (llm-project-workflow):
-- Fail closed: if state is unclear or a step would delete unmerged work, STOP.
+- BLOCKED + DIAGNOSE default: if state is unclear, a required check fails, or a step would delete unmerged work or bypass a guard, STOP and emit a full blocked-diagnose-report.md using the template. Clearly state BLOCKED. Diagnose. Only non-mutating recovery.
 - Never push master. Never discard commits not safely pushed to <remote>.
+- Prove you are in a branches/ worktree before any recovery mutation.
 
 Diagnose first:
 1. git fetch <remote> --prune
@@ -16,8 +19,19 @@ Diagnose first:
 3. git rev-list --left-right --count <remote>/master...master   # ahead/behind
 4. For any PR involved: confirm state (open/closed/merged) AND whether
    <remote>/master actually contains its commits ("closed" != "merged").
+5. Check active leases, claims, and whether required skills/workflows are loaded.
 
-Act per case:
+If a required diagnostic or recovery step itself is unavailable (e.g. terminal broken, skill missing, guard blocks, wrong profile), emit:
+
+## Required step
+<the step>
+
+## Observed failure
+<...>
+
+(complete the full blocked-diagnose-report.md template)
+
+Act per case (only after clean diagnosis; if blocked, use the template and stop):
 - Dirty worktree from another issue: leave it; start yours in a new worktree.
 - Local master ahead of remote: confirm the extra commits live on a branch
   pushed to <remote>, THEN git reset --hard <remote>/master. Verify with
@@ -26,6 +40,7 @@ Act per case:
 - Branch deleted before merge: recover commits from a local branch/reflog (or
   git fsck --lost-found), re-push, reopen the PR.
 - Unauthorized untracked file: do not commit it; leave pre-existing artifacts.
+- Any blocker: use blocked-diagnose-report.md template and stop.
 
-Handoff: what was wrong, evidence, action taken, current state, what remains.
+Handoff: what was wrong, evidence, action taken, current state, what remains. If BLOCKED, include the full blocker report.
 ```

skills/llm-project-workflow/templates/review-pr.md

@@ -36,6 +36,10 @@ Load the canonical workflow first:
 Final report schema: `schemas/review-merge-final-report.md`.
 
 Rules (llm-project-workflow):
+- **BLOCKED + DIAGNOSE default (required):** If any required step (load workflow, lease, profile/role, capability, worktree under branches/, preflight, tool, instruction, etc.) cannot be performed, STOP. State BLOCKED. Use [`blocked-diagnose-report.md`](../templates/blocked-diagnose-report.md) template exactly. Only safe non-mutating recovery. Report. Do not continue or fallback.
+- Find the latest CTH comment on the PR/issue thread before starting work.
+  Post a new CTH: Reviewer Handoff (or CTH: Blocker) at session end.
+  Template: skills/llm-project-workflow/templates/canonical-thread-handoff.md
 - Repository targeting (#530): pass explicit `remote=`, `org=`, and `repo=` on
   every gitea-tools call (e.g. `remote=prgs org=Scaled-Tech-Consulting
   repo=Gitea-Tools`). A bare `remote=prgs` can resolve to the wrong default repo
@@ -110,8 +114,9 @@ Steps:
    - base branch unchanged
    - no undismissed REQUEST_CHANGES / blocking review state left unaccounted
    If anything moved → STOP, re-pin, re-validate before any verdict.
-10. Post the review verdict: approve only if scope is clean and checks pass;
-    otherwise request changes with specifics. Never merge from this review step.
+ 10. Post the review verdict: approve only if scope is clean and checks pass;
+     otherwise request changes with specifics. Never merge from this review step.
+     Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
    Include a "Review Metadata" block (attribution only — docs/llm-agent-sha.md):
 
    Review Metadata:
@@ -128,6 +133,8 @@ Pinned reviewed head, Review decision, Merge result, Linked issue status,
 Cleanup status. If you could not merge, name the exact gate. Reports missing
 the handoff are downgraded (review_proofs.assess_controller_handoff).
 
+Review and merge are separate workflow roles. A reviewer approval is not merge authorization.
+
 Baseline failure proof (#533): if a validation command exits non-zero, do NOT
 call it a clean pass. Only label a failure "baseline"/"pre-existing" with
 pre-merge proof — the failure reproduced on the PR's pre-merge base commit, or a

skills/llm-project-workflow/templates/start-issue.md

@@ -10,6 +10,10 @@ Final report schema: skills/llm-project-workflow/schemas/work-issue-final-report
 Router: skills/llm-project-workflow/SKILL.md (task mode: work-issue)
 
 Rules (llm-project-workflow):
+- **BLOCKED + DIAGNOSE default (required):** If any required step (load workflow, acquire lease, prove worktree under branches/, capability, profile, tool, instruction, preflight, etc.) cannot be performed, STOP. State BLOCKED. Use [`blocked-diagnose-report.md`](../templates/blocked-diagnose-report.md) template exactly. Only safe non-mutating recovery. Report. Do not continue or fallback.
+- Find the latest CTH comment on the issue/PR thread before starting work.
+  Post a new CTH: Author Handoff (or CTH: Blocker) at session end.
+  Template: skills/llm-project-workflow/templates/canonical-thread-handoff.md
 - No repo changes without a tracking issue. If none exists, create one first;
   if it can't be created, stop.
 - Work only in an isolated branch worktree under branches/. The main checkout

skills/llm-project-workflow/workflows/create-issue.md

@@ -12,6 +12,8 @@ This file is the canonical issue-creation workflow for Gitea-Tools. Load it
 before any issue mutation. Final report schema:
 [`schemas/create-issue-final-report.md`](../schemas/create-issue-final-report.md).
 
+**BLOCKED + DIAGNOSE (universal default):** If at any point you cannot perform a required step (skill not available, terminal broken, capability missing, wrong profile, guard blocks, worktree misbound, instructions unavailable, preflight fails, etc.), STOP. Clearly state BLOCKED. Use the standard [`../templates/blocked-diagnose-report.md`](../templates/blocked-diagnose-report.md) template. Only safe non-mutating recovery. Report using the template. Do not continue or use any fallback (temp scripts, direct API, etc.) unless controller authorizes. All blocker classes listed in llm-project-workflow/SKILL.md must trigger this.
+
 **Default task prompt:**
 
 > Create or update Gitea issues in this project only if every identity,
@@ -342,6 +344,41 @@ If edits are needed, make the smallest correction necessary and report the corre
 
 Do not silently change requested meaning.
 
+## 14a. Enforced content preflight (#582)
+
+`gitea_create_issue` runs a **content preflight gate** before duplicate
+search or creation. It fails closed (returns `BLOCKED + DIAGNOSE`, no issue
+created) when the title/body is a *vague reference* to out-of-band draft
+content the session cannot prove it holds, and no durable source pointer is
+present.
+
+An LLM must never fabricate issue content from stale chat memory. If asked to
+create "the drafted issue" / "the prepared issue" / "the issue we discussed"
+/ "the previous draft" without the full content in the current context or a
+durable source pointer, stop and return `BLOCKED + DIAGNOSE` naming the exact
+vague/missing fields.
+
+A **durable source pointer** is one of: an existing issue/PR reference
+(`#582`), a comment id (`comment 8155`), a checked-in file path
+(`task_capability_map.py`), a URL, or a scratchpad path. When a pointer is
+present, retrieve the real content from it before creating.
+
+The gate does **not** require a non-empty body; explicit title-only creation
+stays allowed. It only blocks placeholder references. An operator may bypass
+it with `allow_incomplete_content=True` (report the override).
+
+**Invalid prompts (must block):**
+
+* "Create the drafted issue."
+* "File the prepared issue we discussed."
+* "Open the previous draft."
+
+**Valid prompts (proceed):**
+
+* Full title + body + acceptance criteria supplied inline.
+* "Create the issue drafted in `#582`." (durable pointer → fetch and use it)
+* "Create the issue from `scratchpad/issue-draft.md`." (durable file pointer)
+
 ## 15. Labels, assignees, and metadata
 
 Apply labels, assignees, milestones, or project fields only if:

skills/llm-project-workflow/workflows/review-merge-pr.md

@@ -12,6 +12,10 @@ This file is the canonical PR review/merge workflow for Gitea-Tools. Load it
 before any PR mutation. Final report schema:
 [`schemas/review-merge-final-report.md`](../schemas/review-merge-final-report.md).
 
+**BLOCKED + DIAGNOSE (universal default):** If at any point you cannot perform a required step (skill not available, terminal broken, capability missing, wrong profile, guard blocks, worktree misbound, instructions unavailable, preflight fails, etc.), STOP. Clearly state BLOCKED. Use the standard [`../templates/blocked-diagnose-report.md`](../templates/blocked-diagnose-report.md) template. Only safe non-mutating recovery. Report using the template. Do not continue or use any fallback (temp scripts, direct API, etc.) unless controller authorizes. All blocker classes listed in llm-project-workflow/SKILL.md must trigger this.
+
+**Native MCP failure is a hard stop (#695):** If the native MCP namespace dies (EOF, capability disconnect, session death), STOP. Do **not** import `gitea_mcp_server` from a standalone process, do **not** run offline helpers (`offline_mcp_helper.py`, `offline_mcp_runner.py`, `run_quarantine.py`), and do **not** set direct-import / keychain-bypass / raw-token environment variables. Reconnect the official MCP daemon only. Contaminated approvals must be controller-quarantined; they never authorize merge.
+
 **Default task prompt:**
 
 > Review the next eligible open PR in this project. Merge it only if every
@@ -26,6 +30,12 @@ workflow rules exactly.
 
 ## 0. Load the canonical workflow first
 
+Before starting PR work, **find the latest CTH comment** on the PR or linked
+issue thread. Treat that CTH as the current handoff state. Post a new
+**CTH: Reviewer Handoff**, **CTH: Merger Handoff**, **CTH: Blocker**, or
+**CTH: Supersession Notice** when you finish, block, skip, supersede,
+approve, request changes, or hand off. See `docs/canonical-thread-handoff.md`.
+
 Before starting PR work, check whether the project provides a canonical PR review/merge workflow through a project skill, runbook, or MCP helper.
 
 If available, load it first and report:
@@ -818,6 +828,75 @@ The final report must identify:
 * whether same-PR merge continuation was allowed
 * whether the run stopped as required
 
+## 26A-1. Stale durable review-decision lock cleanup (#594)
+
+After #559, the review-decision lock is durable under
+`~/.cache/gitea-tools/session-state/` (for example
+`review_decision_lock-prgs-reviewer.json`). That durability can outlive the work
+it protects: a terminal `approve` / `request_changes` on a PR that is already
+**merged or closed** still hard-stops later unrelated reviews under #332.
+
+**Do not** delete session-state files by hand. Use the canonical MCP path.
+
+### When cleanup is allowed
+
+Use `gitea_cleanup_stale_review_decision_lock` when:
+
+1. `gitea_mark_final_review_decision` / live review is blocked by
+   `terminal review mutation already consumed ... (#332)`.
+2. Live Gitea state for the **last terminal mutation's PR** is **merged** or
+   **closed** (no same-PR merge sequence remains).
+3. Active profile identity matches the durable lock (reviewer profile with
+   `gitea.pr.review` for `apply=true`).
+4. Optional pin: pass `expected_terminal_pr` to the prior terminal PR number
+   (for example `586` when resuming after a merged #586 lock).
+
+Recommended sequence:
+
+```text
+gitea_whoami
+gitea_resolve_task_capability(task="cleanup_stale_review_decision_lock")
+gitea_cleanup_stale_review_decision_lock(apply=false, remote=..., org=..., repo=...)
+# inspect is_moot / cleanup_allowed / last_terminal_pr
+gitea_cleanup_stale_review_decision_lock(
+    apply=true,
+    expected_terminal_pr=<last_terminal_pr>,
+    remote=..., org=..., repo=...,
+)
+gitea_resolve_task_capability(task="review_pr")
+gitea_load_review_workflow()
+# continue formal mark + submit for the *new* PR
+```
+
+Successful `apply=true` clears memory + durable store and returns an audit
+record (and posts a durable PR comment on the terminal PR when
+`post_audit_comment` is true and comment permission allows).
+
+Same-profile auto-expire: after `gitea_merge_pr` succeeds for the PR that this
+profile just approved, the decision lock for that profile is cleared
+automatically. Cross-profile locks (for example reviewer approve, merger merge)
+still require `gitea_cleanup_stale_review_decision_lock`.
+
+### When cleanup is forbidden
+
+Refuse / fail-closed — keep #332 hard-stop — when:
+
+* the last terminal PR is still **open** (active review or merge sequence may
+  still apply)
+* live PR state cannot be fetched (ambiguous)
+* no lock or no terminal live mutation exists
+* profile identity does not match the lock
+* apply requested without authenticated identity or `gitea.pr.review`
+* `expected_terminal_pr` does not match the last terminal PR
+
+Do **not** use cleanup to:
+
+* bypass #332 on an open PR
+* replace `gitea_authorize_review_correction` for a mistaken review on a still
+  active PR (#211)
+* auto-approve or auto-merge any PR
+* justify `rm` of session-state files
+
 ## 26B. Per-PR reviewer lease (#407)
 
 Parallel reviewer sessions are allowed only when each session holds a distinct,
@@ -928,6 +1007,14 @@ Confirm:
 
 Clean only the session-owned `branches/` review worktree if the project workflow explicitly allows cleanup.
 
+Do not delete source branches from reviewer mode with raw git commands. Commands
+such as `git branch -d <branch>` and `git push <remote> --delete <branch>` are
+not proof of authorized cleanup and bypass MCP role/capability gates. Merged PR
+source branch cleanup must use an explicit MCP cleanup tool such as
+`gitea_cleanup_merged_pr_branch` or another approved cleanup helper with exact
+`gitea.branch.delete` authority, merged-PR proof, no open PR using the branch,
+target-branch ancestry proof, a `branches/` worktree path, and cleanup mutations
+reported separately from review mutations.
 Review, baseline, and merge-simulation worktrees created during this run are
 transient and are removed automatically at successful completion once they are
 clean, carry no open PR, and hold no active lease (#401). Use
@@ -1027,6 +1114,8 @@ Blocked handoffs must say to rerun the full workflow after the blocker clears.
 
 If the blocker is a terminal review mutation already consumed in the current run, the handoff must say that the next run must start from the beginning and must not reuse stale ready/approved state.
 
+If the referenced terminal PR is already **merged or closed** on live Gitea, the durable #332 decision lock is **moot**. Do **not** delete session-state files by hand. Use `gitea_cleanup_stale_review_decision_lock` (assess with `apply=false`, then `apply=true` with `expected_terminal_pr` set) from the reviewer profile after identity/profile checks (#594). Cleanup is **forbidden** while that PR is still open.
+
 ## 30. Final report must be precise
 
 Include:

skills/llm-project-workflow/workflows/work-issue.md

@@ -12,6 +12,10 @@ This file is the canonical author/coder workflow for Gitea-Tools. Load it
 before any issue implementation mutation. Final report schema:
 [`schemas/work-issue-final-report.md`](../schemas/work-issue-final-report.md).
 
+**BLOCKED + DIAGNOSE (universal default):** If at any point you cannot perform a required step (skill not available, terminal broken, capability missing, wrong profile, guard blocks, worktree misbound, instructions unavailable, preflight fails, etc.), STOP. Clearly state BLOCKED. Use the standard [`../templates/blocked-diagnose-report.md`](../templates/blocked-diagnose-report.md) template. Only safe non-mutating recovery. Report using the template. Do not continue or use any fallback (temp scripts, direct API, etc.) unless controller authorizes. All blocker classes listed in llm-project-workflow/SKILL.md must trigger this.
+
+**Native MCP failure is a hard stop (#695):** Do not import MCP server internals, run offline mutation helpers, or set credential-bypass env vars after a native transport failure. Reconnect the official daemon only.
+
 **Default task prompt:**
 
 > Find the next eligible issue in this project, work on it only if all gates
@@ -26,6 +30,12 @@ This is an author/coder workflow. It is not a reviewer workflow.
 
 ## 0. Load the canonical workflow first
 
+Before starting issue work, **find the latest CTH comment** on the target
+issue or linked PR thread. Treat that CTH as the current handoff state unless
+a newer authoritative gate supersedes it. Post a new **CTH: Author Handoff**
+(or `CTH: Blocker`) when you finish, block, or hand off.
+See `docs/canonical-thread-handoff.md`.
+
 Before starting issue work, check whether the project provides a canonical work-on-issue workflow through a project skill, runbook, or MCP helper.
 
 If available, load it first and report:
@@ -615,10 +625,39 @@ After push, report:
 
 If push fails, stop and produce a recovery handoff.
 
-## 20A. Conflict-fix lease and push gate (#399)
+## 20A. Live head re-pin before conflict-fix classification (#522)
+
+Open PR inventory `mergeable` / `head_sha` fields are **advisory and may be
+stale**. Before classifying a PR as conflicted or creating a conflict-fix
+worktree:
+
+1. Re-fetch the live PR (`gitea_view_pr` or equivalent) and record:
+   * inventory head SHA (if any)
+   * inventory mergeable (if any)
+   * **live** head SHA (full 40-char)
+   * **live** mergeable
+2. Call `gitea_assess_conflict_fix_classification` with those values.
+3. Act only on the returned classification and **pinned live head**:
+   * `stale_inventory_skip` / `live_mergeable_skip` → **do not** create a
+     conflict-fix worktree; do not mutate the PR branch for conflicts.
+   * `conflict_fix_needed` → conflict-fix worktree allowed for the pinned
+     live head only.
+   * `incomplete_live_repin` → stop; re-fetch live state.
+4. If inventory head ≠ live head, report both SHAs and use the live head only.
+5. If inventory says `mergeable:false` but live says `mergeable:true`, classify
+   as stale inventory and skip author conflict-fix mutation.
+
+Conflict-fix final reports that claim conflict-fix work must include:
+
+* citation of `gitea_assess_conflict_fix_classification`
+* `Live head SHA: <40-char hex>` (or Pinned head SHA / Live PR head SHA)
+* `Conflict-fix classification: <stale_inventory_skip|conflict_fix_needed|live_mergeable_skip|incomplete_live_repin>`
+
+## 20B. Conflict-fix lease and push gate (#399)
 
 When pushing to an existing PR branch to resolve merge conflicts:
 
+0. Complete §20A live-head re-pin / classification first.
 1. Call `gitea_acquire_conflict_fix_lease` before any push.
 2. Call `gitea_assess_conflict_fix_push` immediately before `git push` with:
    * branch head before push

stable_branch_push_guard.py

@@ -0,0 +1,478 @@
+"""Fail-closed guard against direct pushes to stable branches (#671).
+
+MCP workflow sessions must never publish to a stable branch (``master``,
+``main``, ``dev``, ...) directly. Stable-branch updates land only through
+sanctioned Gitea merge tooling or an explicitly authorized reconciler path.
+
+Incident origin (#670): a bare direct-to-master commit ``2fa97c26`` landed on
+``prgs/master`` without PR/review provenance, and a PR #654 merger run
+attempted ``git push prgs master`` (audited as a no-op, but the *intent* is the
+hazard). Partial protections already existed (``author_proofs.PROTECTED_BRANCHES``,
+``root_checkout_guard``, branch-push proofs) but did not detect shell push
+equivalents, mark the session contaminated after such an attempt, or fail
+closed on subsequent review/merge/close/completion mutations.
+
+This module is the detection + contamination-policy core. Like
+``author_proofs`` and ``root_checkout_guard`` it is **pure**: callers gather
+the raw facts (the proposed command line, git state, the durable contamination
+marker) and pass them in, so the same logic serves prompts, MCP gates, and
+tests. Nothing here performs git or network calls or reads durable state; the
+server wires these helpers to ``mcp_session_state`` and ``verify_preflight_purity``.
+
+Design rules honoured (from the #671 acceptance criteria):
+
+* Detect ``git push <remote> master`` shell equivalents, including refspecs
+  (``HEAD:master``, ``+refs/heads/x:refs/heads/master``), ``--force``,
+  ``--dry-run``/no-op intent, and delete refspecs (``:master``).
+* Never false-block a feature-branch push (``git push prgs fix/issue-671-...``).
+* Never treat ``git fetch`` / ``git pull --ff-only`` as a push.
+* Never treat sanctioned ``gitea_merge_pr`` / Gitea API merge as a push.
+* Fail closed on ambiguous targets that *may* resolve to a stable branch, but
+  surface ambiguity as its own signal rather than silently blocking feature work.
+* Contamination must not be clearable by the same worker session — only a
+  reconciler (audit) role may clear or bypass the gate.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Any, Iterable
+
+from author_proofs import PROTECTED_BRANCHES
+
+# Single source of truth for the stable branch set: the same protected set the
+# author branch-identity proofs already enforce (#177), so the two guards can
+# never drift apart.
+STABLE_BRANCHES = PROTECTED_BRANCHES
+
+# Mutations that must fail closed once the session is contaminated by a direct
+# stable-branch push attempt (#671 AC4: review, merge, close, completion).
+# ``comment_issue`` / ``lock_issue`` / ``create_issue`` are deliberately NOT
+# gated so a contaminated worker can still post the durable audit comment and
+# hand off. Only a reconciler (audit) role bypasses this set.
+CONTAMINATION_GATED_TASKS = frozenset({
+    "create_pr",
+    "commit_files",
+    "gitea_commit_files",
+    "close_pr",
+    "close_issue",
+    "review_pr",
+    "approve_pr",
+    "request_changes_pr",
+    "submit_pr_review",
+    "merge_pr",
+    "delete_branch",
+    "complete_issue",
+})
+
+CONTAMINATION_KIND = "stable_branch_push"
+
+REMEDIATION = (
+    "Direct stable-branch publication is forbidden for worker sessions. Stop, "
+    "leave the stable branch untouched, and route the change through sanctioned "
+    "Gitea merge tooling (gitea_merge_pr) or an explicitly authorized reconciler "
+    "audit. This session is workflow-contaminated until a reconciler audits it."
+)
+
+# ── command tokenising ────────────────────────────────────────────────────────
+
+# Split a compound command line into individual simple commands on shell
+# separators so ``a && git push prgs master`` is analysed segment by segment.
+_SEGMENT_SPLIT_RE = re.compile(r"(?:\|\||&&|\||;|\n)")
+
+_GIT_PUSH_RE = re.compile(r"\bgit\b[\w\s\-]*?\bpush\b", re.IGNORECASE)
+_GIT_FETCH_RE = re.compile(r"\bgit\b[\w\s\-]*?\b(?:fetch|pull)\b", re.IGNORECASE)
+
+# Flags that take a value argument in ``git push`` (so the following token is
+# consumed as the flag's value, not a refspec).
+_VALUE_FLAGS = frozenset({
+    "--repo",
+    "-o",
+    "--push-option",
+    "--receive-pack",
+    "--exec",
+})
+
+_FORCE_FLAGS = frozenset({"-f", "--force"})
+_DRY_RUN_FLAGS = frozenset({"-n", "--dry-run"})
+_DELETE_FLAGS = frozenset({"-d", "--delete"})
+
+
+def _clean(value: str | None) -> str:
+    return (value or "").strip()
+
+
+def _strip_ref_prefix(ref: str) -> str:
+    ref = ref.strip()
+    for prefix in ("refs/heads/", "heads/"):
+        if ref.startswith(prefix):
+            return ref[len(prefix):]
+    return ref
+
+
+def is_stable_ref(ref: str | None) -> bool:
+    """True when ``ref`` names a configured stable branch."""
+    name = _strip_ref_prefix(_clean(ref))
+    return bool(name) and name in STABLE_BRANCHES
+
+
+# ── redaction ─────────────────────────────────────────────────────────────────
+
+_URL_USERINFO_RE = re.compile(r"(https?://)[^/\s:@]+(?::[^/\s@]+)?@", re.IGNORECASE)
+_SECRET_ASSIGN_RE = re.compile(
+    r"\b((?:GITEA_)?(?:TOKEN|PASSWORD|PASS|PAT|SECRET|API_KEY|AUTH))\s*=\s*\S+",
+    re.IGNORECASE,
+)
+_BEARER_RE = re.compile(r"\b(Bearer|token)\s+[A-Za-z0-9._\-]{8,}", re.IGNORECASE)
+
+
+def redact_command(command: str | None) -> str:
+    """Strip credentials/URLs from a command line before logging it (#671).
+
+    Redacts URL userinfo (``https://user:tok@host`` → ``https://***@host``),
+    ``TOKEN=...`` style assignments, and bearer/token headers. Leaves the
+    structural parts (``git push <remote> master``) intact for audit value.
+    """
+    text = _clean(command)
+    if not text:
+        return ""
+    text = _URL_USERINFO_RE.sub(r"\1***@", text)
+    text = _SECRET_ASSIGN_RE.sub(lambda m: f"{m.group(1)}=***", text)
+    text = _BEARER_RE.sub(lambda m: f"{m.group(1)} ***", text)
+    return text
+
+
+# ── push classification ───────────────────────────────────────────────────────
+
+def _analyse_push_segment(segment: str) -> dict[str, Any]:
+    """Classify one ``git push ...`` command segment."""
+    tokens = segment.split()
+    # Drop everything up to and including the ``push`` verb.
+    try:
+        push_idx = next(
+            i for i, tok in enumerate(tokens)
+            if tok.lower() == "push"
+        )
+    except StopIteration:
+        push_idx = -1
+    args = tokens[push_idx + 1:] if push_idx >= 0 else []
+
+    is_force = False
+    is_dry_run = False
+    is_delete = False
+    positionals: list[str] = []
+
+    skip_next = False
+    for tok in args:
+        if skip_next:
+            skip_next = False
+            continue
+        if tok in _FORCE_FLAGS or tok.startswith("--force-with-lease") or tok.startswith("--force-if-includes"):
+            is_force = True
+            continue
+        if tok in _DRY_RUN_FLAGS:
+            is_dry_run = True
+            continue
+        if tok in _DELETE_FLAGS:
+            is_delete = True
+            continue
+        if tok in _VALUE_FLAGS:
+            skip_next = True
+            continue
+        if tok.startswith("--") and "=" in tok:
+            # e.g. --repo=... : self-contained, not a refspec.
+            continue
+        if tok.startswith("-"):
+            # Unknown/combined short flag; ignore for ref detection.
+            continue
+        positionals.append(tok)
+
+    # First positional after push is the remote (when any positional exists);
+    # the rest are refspecs / branch names.
+    remote = positionals[0] if positionals else None
+    refspecs = positionals[1:]
+
+    stable_refs: list[str] = []
+    force_to_stable = False
+    delete_stable = False
+    for spec in refspecs:
+        force_spec = spec.startswith("+")
+        body = spec[1:] if force_spec else spec
+        if ":" in body:
+            src, _, dst = body.partition(":")
+            dest = dst
+            if src == "" and dst:
+                # ``:master`` — delete refspec.
+                is_delete = True
+        else:
+            dest = body
+        if is_stable_ref(dest):
+            stable_refs.append(_strip_ref_prefix(dest))
+            if force_spec or is_force:
+                force_to_stable = True
+            if is_delete:
+                delete_stable = True
+
+    targets_stable = bool(stable_refs)
+    # Ambiguous: ``git push <remote>`` (or bare ``git push``) with no refspec —
+    # resolves to the current branch's upstream, which we cannot see from the
+    # command alone. Flag it, but do not assert stable (would false-block
+    # feature-branch pushes).
+    ambiguous = not refspecs
+
+    return {
+        "is_git_push": True,
+        "remote": remote,
+        "refspecs": refspecs,
+        "targets_stable": targets_stable,
+        "stable_refs": stable_refs,
+        "is_force": is_force or force_to_stable,
+        "is_dry_run": is_dry_run,
+        "is_delete": is_delete or delete_stable,
+        "ambiguous_target": ambiguous,
+    }
+
+
+def classify_push_command(command: str | None) -> dict[str, Any]:
+    """Classify a shell command line for direct stable-branch push intent (#671).
+
+    Returns a dict describing whether the command is a ``git push`` targeting a
+    stable branch. Fetch/pull are never pushes. A sanctioned Gitea merge (which
+    is not a ``git push`` at all) classifies as non-push. Dry-run and no-op
+    pushes still count as *intent* and are reported as contamination
+    (``proves_intent``) so a ``--dry-run`` cannot be used to probe the gate.
+    """
+    text = _clean(command)
+    result: dict[str, Any] = {
+        "command": text,
+        "redacted_command": redact_command(text),
+        "is_git_push": False,
+        "is_fetch_or_pull": False,
+        "targets_stable": False,
+        "stable_refs": [],
+        "is_force": False,
+        "is_dry_run": False,
+        "is_delete": False,
+        "ambiguous_target": False,
+        "contamination": False,
+        "proves_intent": False,
+        "reasons": [],
+    }
+    if not text:
+        return result
+
+    stable_refs: list[str] = []
+    saw_push = False
+    for segment in _SEGMENT_SPLIT_RE.split(text):
+        seg = segment.strip()
+        if not seg:
+            continue
+        if _GIT_FETCH_RE.search(seg) and not _GIT_PUSH_RE.search(seg):
+            result["is_fetch_or_pull"] = True
+            continue
+        if not _GIT_PUSH_RE.search(seg):
+            continue
+        saw_push = True
+        analysis = _analyse_push_segment(seg)
+        result["is_git_push"] = True
+        result["is_force"] = result["is_force"] or analysis["is_force"]
+        result["is_dry_run"] = result["is_dry_run"] or analysis["is_dry_run"]
+        result["is_delete"] = result["is_delete"] or analysis["is_delete"]
+        result["ambiguous_target"] = result["ambiguous_target"] or analysis["ambiguous_target"]
+        stable_refs.extend(analysis["stable_refs"])
+
+    result["stable_refs"] = sorted(set(stable_refs))
+    result["targets_stable"] = bool(stable_refs)
+
+    reasons: list[str] = []
+    if result["targets_stable"]:
+        refs = ", ".join(result["stable_refs"])
+        verb = "delete" if result["is_delete"] else ("force-push" if result["is_force"] else "push")
+        qualifier = " (dry-run/no-op still proves intent)" if result["is_dry_run"] else ""
+        reasons.append(
+            f"direct stable-branch {verb} detected targeting: {refs}{qualifier}; "
+            "worker sessions must never publish stable branches directly"
+        )
+        result["contamination"] = True
+        result["proves_intent"] = True
+    elif saw_push and result["ambiguous_target"]:
+        # Bare ``git push`` with no refspec: ambiguous. Report, but do not
+        # contaminate on the command alone — the root-checkout / branch-state
+        # detector resolves whether the current branch is stable.
+        reasons.append(
+            "ambiguous 'git push' with no refspec: resolve the current branch "
+            "before pushing; if it is a stable branch this is forbidden"
+        )
+
+    result["reasons"] = reasons
+    return result
+
+
+def detect_stable_push(commands: str | Iterable[str] | None) -> dict[str, Any]:
+    """Classify one command or an iterable of commands; contamination if any hit."""
+    if commands is None:
+        return classify_push_command(None)
+    if isinstance(commands, str):
+        return classify_push_command(commands)
+    worst: dict[str, Any] | None = None
+    for cmd in commands:
+        res = classify_push_command(cmd)
+        if res["contamination"]:
+            return res
+        if worst is None or (res["is_git_push"] and not worst["is_git_push"]):
+            worst = res
+    return worst or classify_push_command(None)
+
+
+# ── root/control checkout local-commit detection ──────────────────────────────
+
+def assess_root_checkout_local_commit(
+    *,
+    current_branch: str | None,
+    head_sha: str | None,
+    remote_master_sha: str | None,
+    is_under_branches: bool,
+    ahead_count: int | None = None,
+) -> dict[str, Any]:
+    """Detect a local commit on the root/control checkout not on a feature branch.
+
+    #671 AC2. An isolated ``branches/...`` worktree is exempt (that is where
+    feature work belongs). On the control checkout, a commit is illegitimate
+    when the checkout sits on a stable branch (or detached) and its HEAD has
+    advanced past the tracking ``prgs/master`` — i.e. a commit was made that is
+    not carried by an issue feature branch/PR.
+
+    Positive evidence only: missing state reports ``unknown`` rather than
+    asserting contamination, but an unreadable *branch* on the control checkout
+    (detached HEAD) with an advanced HEAD is still flagged.
+    """
+    branch = _clean(current_branch)
+    head = _clean(head_sha).lower()
+    master = _clean(remote_master_sha).lower()
+
+    if is_under_branches:
+        return {
+            "contamination": False,
+            "unknown": False,
+            "reasons": [],
+            "detail": "isolated branches/ worktree — feature commits are expected here",
+        }
+
+    reasons: list[str] = []
+    unknown = False
+
+    on_stable = (not branch) or (branch in STABLE_BRANCHES)
+    if not on_stable:
+        # Control checkout is on some non-stable branch: out of scope for this
+        # detector (root_checkout_guard #475 handles that contamination class).
+        return {
+            "contamination": False,
+            "unknown": False,
+            "reasons": [],
+            "detail": f"control checkout on non-stable branch '{branch}' — not this detector's class",
+        }
+
+    advanced = False
+    if ahead_count is not None and ahead_count > 0:
+        advanced = True
+    if head and master and head != master:
+        advanced = True
+    if (not head or not master) and ahead_count is None:
+        unknown = True
+
+    if advanced:
+        where = f"branch '{branch}'" if branch else "detached HEAD"
+        reasons.append(
+            f"control checkout ({where}) has local commits not on an issue "
+            "feature branch (HEAD advanced past prgs/master); worker sessions "
+            "must commit only on issue branches under branches/"
+        )
+
+    return {
+        "contamination": bool(reasons),
+        "unknown": unknown and not reasons,
+        "reasons": reasons,
+        "current_branch": branch or None,
+        "head_sha": head or None,
+        "remote_master_sha": master or None,
+    }
+
+
+# ── contamination record + gate ───────────────────────────────────────────────
+
+def build_contamination_record(
+    *,
+    reason_class: str,
+    command_redacted: str | None = None,
+    session_id: str | None = None,
+    remote: str | None = None,
+    ref: str | None = None,
+    role: str | None = None,
+    detail: str | None = None,
+) -> dict[str, Any]:
+    """Build the durable contamination marker payload (redacted, audit-safe).
+
+    ``reason_class`` is one of ``stable_branch_push`` / ``root_checkout_commit``.
+    The command is stored already-redacted; callers pass the raw line through
+    :func:`redact_command` (or this builder redacts a raw ``command`` for them).
+    """
+    return {
+        "kind": CONTAMINATION_KIND,
+        "reason_class": _clean(reason_class) or "stable_branch_push",
+        "command_summary": redact_command(command_redacted),
+        "session_id": _clean(session_id) or None,
+        "remote": _clean(remote) or None,
+        "ref": _clean(ref) or None,
+        "role": _clean(role) or None,
+        "detail": _clean(detail) or None,
+        "cleared_by_reconciler": False,
+    }
+
+
+def assess_contamination_gate(
+    marker: dict[str, Any] | None,
+    *,
+    task: str | None,
+    actual_role: str | None,
+) -> dict[str, Any]:
+    """Fail closed on gated mutations while a contamination marker is live (#671 AC4).
+
+    * No marker → allowed.
+    * Reconciler (audit) role → allowed (the sanctioned path to inspect/clear).
+    * Marker present + ``task`` in :data:`CONTAMINATION_GATED_TASKS` → blocked.
+    * Marker present + non-gated task (e.g. ``comment_issue``) → allowed, so the
+      worker can still post the durable audit/handoff comment.
+    """
+    if not marker or marker.get("cleared_by_reconciler"):
+        return {"block": False, "reasons": [], "task": task}
+
+    role = _clean(actual_role).lower()
+    if role == "reconciler":
+        return {
+            "block": False,
+            "reasons": [],
+            "task": task,
+            "detail": "reconciler audit path is exempt from the contamination gate",
+        }
+
+    task_name = _clean(task)
+    if task_name and task_name in CONTAMINATION_GATED_TASKS:
+        summary = marker.get("command_summary") or marker.get("detail") or "(no summary)"
+        reason_class = marker.get("reason_class") or "stable_branch_push"
+        return {
+            "block": True,
+            "reasons": [
+                f"session is workflow-contaminated ({reason_class}): {summary}. "
+                f"'{task_name}' is blocked until a reconciler audits and clears "
+                "the contamination. " + REMEDIATION
+            ],
+            "task": task_name,
+        }
+
+    return {"block": False, "reasons": [], "task": task_name or None}
+
+
+def format_contamination_gate_error(gate: dict[str, Any]) -> str:
+    """Single RuntimeError message for MCP mutation gates."""
+    reasons = "; ".join(gate.get("reasons") or ["session workflow-contaminated"])
+    return f"Stable-branch contamination gate (#671): {reasons}"

stale_review_decision_lock.py

@@ -0,0 +1,689 @@
+"""Stale / moot #332 review-decision lock detection and cleanup policy (#594/#620).
+
+#332 correctly hard-stops a reviewer session after a terminal live review
+mutation. After #559 those locks are durable on disk, so they can outlive the
+work they protect: when the referenced PR is already merged or closed, no
+same-PR merge sequence remains, yet the durable ledger still blocks unrelated
+new terminal reviews.
+
+#620 scopes the terminal boundary by **reviewed head SHA**: a prior
+REQUEST_CHANGES (or APPROVE) on head A must not block a fresh formal decision
+on head B of the **same open PR**. Same-head duplicates remain fail-closed.
+Open-PR lock *cleanup* remains forbidden unless the PR is truly merged/closed
+(#594); new-head re-review is allowed without deleting the durable lock.
+
+This module is pure policy (no Gitea I/O). The MCP tool
+``gitea_cleanup_stale_review_decision_lock`` fetches live PR state, calls these
+helpers, and only then clears durable state when cleanup is allowed.
+
+Manual deletion of ``~/.cache/gitea-tools/session-state/*`` is never the
+canonical path.
+"""
+
+from __future__ import annotations
+
+from datetime import datetime, timezone
+from typing import Any
+
+# Keep in sync with gitea_mcp_server._TERMINAL_REVIEW_ACTIONS.
+TERMINAL_REVIEW_ACTIONS = frozenset({"approve", "request_changes"})
+
+
+def normalize_head_sha(value: Any) -> str | None:
+    """Normalize a git SHA for equality comparison; None if empty/invalid."""
+    if value is None:
+        return None
+    text = str(value).strip().lower()
+    if not text:
+        return None
+    return text
+
+
+def heads_equal(a: Any, b: Any) -> bool:
+    na = normalize_head_sha(a)
+    nb = normalize_head_sha(b)
+    if not na or not nb:
+        return False
+    return na == nb
+
+
+def mutation_head_sha(mutation: dict | None, lock: dict | None = None) -> str | None:
+    """Head SHA that bounds a live mutation (#620).
+
+    Prefers fields recorded on the mutation itself. For *legacy* mutations
+    that predate head-scoping, falls back to the lock's
+    ``ready_expected_head_sha`` only when that ready binding is for the same
+    PR number (the frozen durable PR #619-style ledger).
+    """
+    if not isinstance(mutation, dict):
+        return None
+    for key in ("head_sha", "expected_head_sha", "reviewed_head_sha"):
+        head = normalize_head_sha(mutation.get(key))
+        if head:
+            return head
+    if not isinstance(lock, dict):
+        return None
+    if lock.get("ready_pr_number") != mutation.get("pr_number"):
+        return None
+    return normalize_head_sha(lock.get("ready_expected_head_sha"))
+
+
+def last_terminal_mutation(lock: dict | None) -> dict | None:
+    """Return the last terminal live mutation on *lock*, or None."""
+    if not lock:
+        return None
+    terminals = [
+        m
+        for m in (lock.get("live_mutations") or [])
+        if isinstance(m, dict) and m.get("action") in TERMINAL_REVIEW_ACTIONS
+    ]
+    return terminals[-1] if terminals else None
+
+
+def prior_live_mutations_block_boundary(
+    lock: dict | None,
+    *,
+    pr_number: int,
+    expected_head_sha: str | None,
+) -> bool:
+    """True when prior live mutations block a new decision for PR+head (#620).
+
+    Historical terminals on a **different head of the same PR** do not block.
+    Any prior mutation on another PR, the same head, or without a comparable
+    head SHA fails closed (blocks).
+
+    Head resolution uses ``mutation_head_sha(m, lock)`` so pre-#620 ledgers
+    that only stored ``ready_expected_head_sha`` still compare correctly
+    (PR #619-style).
+    """
+    if not lock:
+        return False
+    if lock.get("correction_authorized"):
+        return False
+    prior = list(lock.get("live_mutations") or [])
+    if not prior:
+        return False
+    target = normalize_head_sha(expected_head_sha)
+    for m in prior:
+        if not isinstance(m, dict):
+            return True
+        m_pr = m.get("pr_number")
+        m_head = mutation_head_sha(m, lock)
+        if (
+            m_pr == pr_number
+            and target
+            and m_head
+            and not heads_equal(m_head, target)
+        ):
+            # Same open PR, different reviewed head — historical boundary only.
+            continue
+        return True
+    return False
+
+
+def terminal_boundary_allows_fresh_decision(
+    lock: dict | None,
+    *,
+    pr_number: int,
+    expected_head_sha: str | None,
+    operation: str,
+) -> bool:
+    """Whether #332 hard-stop should yield for a new PR+head boundary (#620).
+
+    Only for ``mark_ready`` / ``review`` / ``resume`` on the **same PR** when
+    the requested head differs from the last terminal's head. Merge is never
+    reopened by head change (approved head merge path is separate).
+    """
+    if operation not in ("mark_ready", "review", "resume"):
+        return False
+    if not lock:
+        return False
+    if lock.get("correction_authorized"):
+        return True
+    last = last_terminal_mutation(lock)
+    if last is None:
+        return True
+    if last.get("pr_number") != pr_number:
+        return False
+    locked_head = mutation_head_sha(last, lock)
+    target = normalize_head_sha(expected_head_sha)
+    if not locked_head or not target:
+        return False
+    return not heads_equal(locked_head, target)
+
+
+def backfill_terminal_heads_from_ready(lock: dict) -> dict:
+    """Stamp head_sha onto legacy terminal mutations before overwriting ready_*.
+
+    Called when marking a fresh decision on a new head so historical rows keep
+    their original boundary after ``ready_expected_head_sha`` advances (#620).
+    """
+    if not isinstance(lock, dict):
+        return lock
+    ready_pr = lock.get("ready_pr_number")
+    ready_head = normalize_head_sha(lock.get("ready_expected_head_sha"))
+    if ready_pr is None or not ready_head:
+        return lock
+    mutations = list(lock.get("live_mutations") or [])
+    changed = False
+    for m in mutations:
+        if not isinstance(m, dict):
+            continue
+        if m.get("action") not in TERMINAL_REVIEW_ACTIONS:
+            continue
+        if m.get("pr_number") != ready_pr:
+            continue
+        if mutation_head_sha(m):
+            continue
+        m["head_sha"] = ready_head
+        changed = True
+    if changed:
+        lock["live_mutations"] = mutations
+    return lock
+
+
+def classify_pr_live_state(pr_live: dict | None) -> dict[str, Any]:
+    """Normalize a Gitea PR payload into merge/closed flags."""
+    pr = pr_live or {}
+    merged = bool(pr.get("merged") or pr.get("merged_at"))
+    state = (pr.get("state") or "").strip().lower() or None
+    closed = state == "closed"
+    head = pr.get("head") if isinstance(pr.get("head"), dict) else {}
+    head_sha = normalize_head_sha(
+        pr.get("head_sha") or pr.get("head_commit_sha") or head.get("sha")
+    )
+    return {
+        "pr_state": state,
+        "pr_merged": merged,
+        "pr_merged_or_closed": merged or closed,
+        "merge_commit_sha": pr.get("merge_commit_sha")
+        or pr.get("merged_commit_sha"),
+        "current_pr_head_sha": head_sha,
+    }
+
+
+def lock_summary(lock: dict | None) -> dict[str, Any] | None:
+    """LLM-safe summary of a decision lock (no secrets)."""
+    if lock is None:
+        return None
+    last = last_terminal_mutation(lock)
+    mutations = list(lock.get("live_mutations") or [])
+    last_head = mutation_head_sha(last, lock) if last else None
+    return {
+        "task": lock.get("task"),
+        "remote": lock.get("remote"),
+        "org": lock.get("org") or lock.get("ready_org"),
+        "repo": lock.get("repo") or lock.get("ready_repo"),
+        "profile_identity": lock.get("profile_identity")
+        or lock.get("session_profile_lock")
+        or lock.get("session_profile"),
+        "session_profile": lock.get("session_profile"),
+        "ready_pr_number": lock.get("ready_pr_number"),
+        "ready_action": lock.get("ready_action"),
+        "ready_expected_head_sha": normalize_head_sha(
+            lock.get("ready_expected_head_sha")
+        ),
+        "live_mutations_count": len(mutations),
+        "last_terminal": (
+            {
+                "pr_number": last.get("pr_number"),
+                "action": last.get("action"),
+                "review_id": last.get("review_id"),
+                "head_sha": last_head,
+            }
+            if last
+            else None
+        ),
+        "locked_head_sha": last_head,
+        "correction_authorized": bool(lock.get("correction_authorized")),
+        "updated_at": lock.get("updated_at") or lock.get("recorded_at"),
+    }
+
+
+def assess_stale_review_decision_lock(
+    lock: dict | None,
+    *,
+    pr_live: dict | None = None,
+    pr_lookup_error: str | None = None,
+    active_profile_identity: str | None = None,
+    current_pr_head_sha: str | None = None,
+) -> dict[str, Any]:
+    """Assess whether a durable review-decision lock is stale/moot (#594/#620).
+
+    *pr_live* must be the live Gitea payload for the **last terminal
+    mutation's PR**. When no lock / no terminal mutation exists, cleanup is
+    not applicable. When the PR is still open (or lookup fails), cleanup is
+    forbidden (#594). Open PR + different live head reports
+    ``stale_by_head`` / ``fresh_review_on_current_head_allowed`` (#620)
+    without enabling cleanup.
+    """
+    summary = lock_summary(lock)
+    result: dict[str, Any] = {
+        "has_lock": lock is not None,
+        "lock_summary": summary,
+        "last_terminal_pr": None,
+        "last_terminal_action": None,
+        "locked_head_sha": None,
+        "current_pr_head_sha": normalize_head_sha(current_pr_head_sha),
+        "stale_by_head": False,
+        "fresh_review_on_current_head_allowed": False,
+        "is_moot": False,
+        "cleanup_allowed": False,
+        "profile_match": True,
+        "pr_state": None,
+        "pr_merged": False,
+        "pr_merged_or_closed": False,
+        "merge_commit_sha": None,
+        "reasons": [],
+        "recovery_tool": "gitea_cleanup_stale_review_decision_lock",
+    }
+
+    if lock is None:
+        result["reasons"].append("no review decision lock present")
+        return result
+
+    # Profile identity gate (caller may re-check with env; pure assess still
+    # reports mismatch when active identity is supplied).
+    stored = (
+        (lock.get("profile_identity") or lock.get("session_profile_lock") or "")
+        .strip()
+    )
+    active = (active_profile_identity or "").strip()
+    if active and stored and active != stored:
+        result["profile_match"] = False
+        result["reasons"].append(
+            "review decision lock profile identity does not match active "
+            f"profile '{active}' (fail closed, #594)"
+        )
+        return result
+
+    last = last_terminal_mutation(lock)
+    if last is None:
+        result["reasons"].append(
+            "review decision lock has no terminal live mutations; nothing to clear"
+        )
+        return result
+
+    pr_number = last.get("pr_number")
+    action = last.get("action")
+    locked_head = mutation_head_sha(last, lock)
+    result["last_terminal_pr"] = pr_number
+    result["last_terminal_action"] = action
+    result["locked_head_sha"] = locked_head
+
+    if pr_number is None:
+        result["reasons"].append(
+            "last terminal mutation missing pr_number; refuse cleanup (fail closed)"
+        )
+        return result
+
+    if pr_lookup_error:
+        result["reasons"].append(
+            f"could not load live state for PR #{pr_number}: {pr_lookup_error} "
+            "(fail closed, #594)"
+        )
+        return result
+
+    if pr_live is None:
+        result["reasons"].append(
+            f"live PR state required for terminal PR #{pr_number} before cleanup "
+            "(fail closed, #594)"
+        )
+        return result
+
+    live = classify_pr_live_state(pr_live)
+    current_head = normalize_head_sha(
+        current_pr_head_sha or live.get("current_pr_head_sha")
+    )
+    result.update(
+        {
+            "pr_state": live["pr_state"],
+            "pr_merged": live["pr_merged"],
+            "pr_merged_or_closed": live["pr_merged_or_closed"],
+            "merge_commit_sha": live["merge_commit_sha"],
+            "current_pr_head_sha": current_head,
+        }
+    )
+
+    if not live["pr_merged_or_closed"]:
+        stale_by_head = bool(
+            locked_head and current_head and not heads_equal(locked_head, current_head)
+        )
+        result["stale_by_head"] = stale_by_head
+        # Fresh formal decision is allowed on the new head without cleanup (#620).
+        result["fresh_review_on_current_head_allowed"] = stale_by_head
+        if stale_by_head:
+            result["reasons"].append(
+                f"terminal {action} on PR #{pr_number} is bound to head "
+                f"{locked_head[:12]}…; live head is {current_head[:12]}… — "
+                "fresh formal review on current head is allowed without "
+                "cleanup (fail closed for cleanup, #620); #332 same-head "
+                "duplicate protection remains"
+            )
+        else:
+            result["reasons"].append(
+                f"terminal review mutation on open PR #{pr_number} is still active "
+                f"({action}); #332 hard-stop remains — cleanup forbidden (#594)"
+            )
+        return result
+
+    # Merged or closed: lock is moot. Same-PR merge continuation is impossible.
+    result["is_moot"] = True
+    result["cleanup_allowed"] = True
+    result["stale_by_head"] = False
+    result["fresh_review_on_current_head_allowed"] = False
+    result["reasons"].append(
+        f"terminal mutation ({action} on PR #{pr_number}) is moot: PR is "
+        f"{'merged' if live['pr_merged'] else 'closed'}; canonical cleanup allowed (#594)"
+    )
+    return result
+
+
+def build_cleanup_audit_record(
+    *,
+    assessment: dict[str, Any],
+    actor_username: str | None,
+    profile_name: str | None,
+    applied: bool,
+) -> dict[str, Any]:
+    """Structured durable audit payload for a cleanup attempt."""
+    last = (assessment.get("lock_summary") or {}).get("last_terminal") or {}
+    return {
+        "event": "stale_review_decision_lock_cleanup",
+        "issue_ref": "#594",
+        "applied": bool(applied),
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+        "actor_username": actor_username,
+        "profile_name": profile_name,
+        "last_terminal_pr": assessment.get("last_terminal_pr") or last.get("pr_number"),
+        "last_terminal_action": assessment.get("last_terminal_action")
+        or last.get("action"),
+        "pr_state": assessment.get("pr_state"),
+        "pr_merged": assessment.get("pr_merged"),
+        "pr_merged_or_closed": assessment.get("pr_merged_or_closed"),
+        "merge_commit_sha": assessment.get("merge_commit_sha"),
+        "prior_live_mutations_count": (
+            (assessment.get("lock_summary") or {}).get("live_mutations_count")
+        ),
+        "prior_profile_identity": (
+            (assessment.get("lock_summary") or {}).get("profile_identity")
+        ),
+        "cleanup_allowed": assessment.get("cleanup_allowed"),
+        "is_moot": assessment.get("is_moot"),
+        "reasons": list(assessment.get("reasons") or []),
+    }
+
+
+def format_cleanup_audit_comment(audit: dict[str, Any]) -> str:
+    """Markdown body for a durable Gitea audit comment after cleanup."""
+    status = "APPLIED" if audit.get("applied") else "ASSESSED (not applied)"
+    lines = [
+        "## Stale #332 review-decision lock cleanup (#594)",
+        "",
+        f"Status: **{status}**",
+        "",
+        f"- actor: `{audit.get('actor_username')}`",
+        f"- profile: `{audit.get('profile_name')}`",
+        f"- timestamp: `{audit.get('timestamp')}`",
+        f"- last terminal: `{audit.get('last_terminal_action')}` on "
+        f"PR #{audit.get('last_terminal_pr')}",
+        f"- PR state: `{audit.get('pr_state')}` "
+        f"(merged={audit.get('pr_merged')})",
+        f"- merge_commit_sha: `{audit.get('merge_commit_sha')}`",
+        f"- prior live_mutations_count: "
+        f"`{audit.get('prior_live_mutations_count')}`",
+        f"- prior profile_identity: `{audit.get('prior_profile_identity')}`",
+        "",
+        "Manual deletion of session-state files is **not** the workflow.",
+        "This path only clears a lock when the referenced PR is merged/closed.",
+    ]
+    return "\n".join(lines)
+
+
+# ---------------------------------------------------------------------------
+# #709 cross-profile cleanup, overwrite protection, recovery provenance
+# ---------------------------------------------------------------------------
+
+
+def has_unresolved_terminal_evidence(lock: dict | None) -> bool:
+    """True when *lock* still carries a terminal live mutation ledger."""
+    return last_terminal_mutation(lock) is not None
+
+
+def assess_init_overwrite(
+    existing_lock: dict | None,
+    *,
+    force: bool = False,
+) -> dict[str, Any]:
+    """Whether init_review_decision_lock may replace an existing durable lock (#709 AC2).
+
+    Unresolved terminal evidence must never be silently replaced by an empty
+    initialized lock. *force* does not authorize destruction of terminal
+    ledgers — only explicit cleanup / archive transitions may remove them.
+    """
+    result: dict[str, Any] = {
+        "overwrite_allowed": True,
+        "has_existing": existing_lock is not None,
+        "has_unresolved_terminal": False,
+        "reasons": [],
+        "last_terminal_pr": None,
+        "last_terminal_action": None,
+        "existing_profile_identity": None,
+    }
+    if existing_lock is None:
+        result["reasons"].append("no existing lock; empty init allowed")
+        return result
+    result["existing_profile_identity"] = (
+        existing_lock.get("profile_identity")
+        or existing_lock.get("session_profile_lock")
+        or existing_lock.get("session_profile")
+    )
+    last = last_terminal_mutation(existing_lock)
+    if last is None:
+        result["reasons"].append(
+            "existing lock has no terminal mutations; re-init allowed"
+        )
+        return result
+    result["has_unresolved_terminal"] = True
+    result["overwrite_allowed"] = False
+    result["last_terminal_pr"] = last.get("pr_number")
+    result["last_terminal_action"] = last.get("action")
+    result["reasons"].append(
+        "refuse empty re-init: unresolved terminal decision-lock evidence "
+        f"present ({last.get('action')} on PR #{last.get('pr_number')}); "
+        "use gitea_cleanup_stale_review_decision_lock when moot, or archive "
+        f"via sanctioned recovery — force={force!r} does not authorize overwrite "
+        "(#709 AC2)"
+    )
+    return result
+
+
+def lock_targets_merged_pr_approval(
+    lock: dict | None,
+    *,
+    pr_number: int,
+    expected_head_sha: str | None = None,
+) -> bool:
+    """True when *lock*'s last terminal mutation is approve of *pr_number*.
+
+    When *expected_head_sha* is provided, a **recorded** terminal head must
+    match. Legacy same-repo approve locks with no recorded head do **not**
+    match (fail closed, #709 F3 residual / review 435) — callers must use the
+    strict secondary path that refuses no-head clears.
+    """
+    last = last_terminal_mutation(lock)
+    if last is None:
+        return False
+    if last.get("action") != "approve":
+        return False
+    if last.get("pr_number") != pr_number:
+        return False
+    if expected_head_sha:
+        want = normalize_head_sha(expected_head_sha)
+        if not want:
+            return False
+        locked = mutation_head_sha(last, lock)
+        # Require recorded-head match; unrecorded head is not a match (#709 F3).
+        if not locked or not heads_equal(locked, want):
+            return False
+    return True
+
+
+def build_post_merge_recovery_record(
+    *,
+    pr_number: int,
+    head_sha: str | None,
+    merge_commit_sha: str | None,
+    target_profile_identity: str | None,
+    failed_step: str,
+    error: str | None,
+    remote: str | None,
+    org: str | None,
+    repo: str | None,
+    actor_username: str | None,
+    profile_name: str | None,
+) -> dict[str, Any]:
+    """Durable recovery-required payload after irreversible merge (#709 AC3)."""
+    return {
+        "event": "post_merge_decision_lock_recovery_required",
+        "status": "recovery_required",
+        "issue_ref": "#709",
+        "recovery_critical": True,
+        "applied": False,  # never claim historical cleanup succeeded
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+        "pr_number": pr_number,
+        "head_sha": normalize_head_sha(head_sha),
+        "merge_commit_sha": merge_commit_sha,
+        "target_profile_identity": target_profile_identity,
+        "failed_step": failed_step,
+        "error": error,
+        "remote": remote,
+        "org": org,
+        "repo": repo,
+        "actor_username": actor_username,
+        "profile_name": profile_name,
+        "required_recovery_action": (
+            "retry cross-profile decision-lock cleanup and audit publication "
+            "for the merged PR; if terminal evidence is gone, use "
+            "gitea_record_irrecoverable_decision_lock_provenance"
+        ),
+    }
+
+
+def build_irrecoverable_provenance_record(
+    *,
+    pr_number: int,
+    head_sha: str | None,
+    remote: str | None,
+    org: str | None,
+    repo: str | None,
+    actor_username: str | None,
+    profile_name: str | None,
+    reason: str,
+    incident_ref: str | None = None,
+    # Deprecated kwargs retained only so stale call sites fail closed:
+    operator_authorized: bool | None = None,
+    # Required for merger-acceptable records (#709 F1):
+    authorization: dict[str, Any] | None = None,
+    incident_issue: int | None = None,
+    incident_comment_id: int | None = None,
+    destroyed_subject: str | None = None,
+    historical_provenance_subject: str | None = None,
+) -> dict[str, Any]:
+    """Truthful absence-of-proof record (#709 AC5). Never sets applied=True.
+
+    Caller-supplied ``operator_authorized`` is **ignored** as authorization
+    evidence (review 434 F1). Prefer
+    :func:`irrecoverable_provenance.build_irrecoverable_provenance_record`
+    with a verified server-side authorization artifact.
+    """
+    # Explicitly ignore deprecated self-assertable Boolean.
+    _ = operator_authorized
+    if authorization is not None and incident_issue is not None and incident_comment_id is not None:
+        from irrecoverable_provenance import (
+            build_irrecoverable_provenance_record as _build,
+        )
+
+        return _build(
+            pr_number=int(pr_number),
+            head_sha=str(head_sha or ""),
+            remote=str(remote or ""),
+            org=str(org or ""),
+            repo=str(repo or ""),
+            actor_username=actor_username,
+            profile_name=profile_name,
+            reason=reason,
+            incident_issue=int(incident_issue),
+            incident_comment_id=int(incident_comment_id),
+            authorization=authorization,
+            destroyed_subject=destroyed_subject,
+            historical_provenance_subject=historical_provenance_subject
+            or destroyed_subject,
+        )
+    # Fail-closed skeleton when no server authorization is supplied: never
+    # sets merger_may_accept True (even if operator_authorized was True).
+    return {
+        "event": "irrecoverable_decision_lock_provenance",
+        "status": "provenance_irrecoverable",
+        "record_type": "irrecoverable_decision_provenance",
+        "operator_recovery_required": True,
+        "issue_ref": "#709",
+        "recovery_critical": True,
+        "applied": False,
+        "historical_cleanup_proven": False,
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+        "pr_number": pr_number,
+        "head_sha": normalize_head_sha(head_sha),
+        "remote": remote,
+        "org": org,
+        "repo": repo,
+        "actor_username": actor_username,
+        "profile_name": profile_name,
+        "reason": reason,
+        "incident_ref": incident_ref,
+        "incident_issue": incident_issue,
+        "incident_comment_id": incident_comment_id,
+        "authorization_verified": False,
+        "merger_may_accept": False,
+        "acceptance_rule": (
+            "Merger may accept this record only when a server-side "
+            "authorization artifact verifies for remote/org/repo/PR/exact "
+            "head/incident, the record is durable and read back, and normal "
+            "merge gates still pass. Caller Booleans never authorize. This "
+            "does not prove historical cleanup."
+        ),
+    }
+
+
+def format_irrecoverable_audit_comment(record: dict[str, Any]) -> str:
+    """Markdown body for irrecoverable provenance audit (no applied=true claim)."""
+    from irrecoverable_provenance import (
+        format_irrecoverable_audit_comment as _fmt,
+    )
+
+    return _fmt(record)
+
+
+def format_post_merge_recovery_comment(record: dict[str, Any]) -> str:
+    """Markdown body for post-merge recovery-required audit."""
+    lines = [
+        "## Post-merge decision-lock recovery required (#709)",
+        "",
+        "Status: **RECOVERY_REQUIRED** (merge is irreversible; cleanup/audit incomplete)",
+        "",
+        f"- actor: `{record.get('actor_username')}`",
+        f"- profile: `{record.get('profile_name')}`",
+        f"- timestamp: `{record.get('timestamp')}`",
+        f"- PR: `#{record.get('pr_number')}`",
+        f"- head_sha: `{record.get('head_sha')}`",
+        f"- merge_commit_sha: `{record.get('merge_commit_sha')}`",
+        f"- target_profile_identity: `{record.get('target_profile_identity')}`",
+        f"- failed_step: `{record.get('failed_step')}`",
+        f"- error: `{record.get('error')}`",
+        "",
+        f"Required action: {record.get('required_recovery_action')}",
+        "",
+        "applied=false — do not treat this as successful cleanup evidence.",
+    ]
+    return "\n".join(lines)
+

task_capability_map.py

@@ -36,6 +36,10 @@ TASK_CAPABILITY_MAP: dict[str, dict[str, str]] = {
         "permission": "gitea.issue.comment",
         "role": "author",
     },
+    "create_label": {
+        "permission": "gitea.issue.comment",
+        "role": "author",
+    },
     "create_branch": {
         "permission": "gitea.branch.create",
         "role": "author",
@@ -66,9 +70,29 @@ TASK_CAPABILITY_MAP: dict[str, dict[str, str]] = {
     },
     "merge_pr": {
         "permission": "gitea.pr.merge",
-        "role": "reviewer",
+        "role": "merger",
+    },
+    # #695 AC8: controller quarantine of contaminated formal reviews.
+    # Apply path posts an append-only forensic audit comment (pr.comment).
+    "quarantine_contaminated_review": {
+        "permission": "gitea.pr.comment",
+        "role": "reconciler",
+    },
+    "gitea_quarantine_contaminated_review": {
+        "permission": "gitea.pr.comment",
+        "role": "reconciler",
     },
     "adopt_merger_pr_lease": {
+        "permission": "gitea.pr.comment",
+        "role": "merger",
+    },
+    # #691: guarded non-owner cleanup of obsolete comment-backed reviewer leases.
+    # Apply path posts lease release + audit comments (gitea.pr.comment).
+    "cleanup_obsolete_reviewer_comment_lease": {
+        "permission": "gitea.pr.comment",
+        "role": "reviewer",
+    },
+    "gitea_cleanup_obsolete_reviewer_comment_lease": {
         "permission": "gitea.pr.comment",
         "role": "reviewer",
     },
@@ -92,10 +116,51 @@ TASK_CAPABILITY_MAP: dict[str, dict[str, str]] = {
         "permission": "gitea.pr.approve",
         "role": "reviewer",
     },
+    # #594: clear durable #332 decision lock only when last terminal PR is
+    # already merged/closed (moot). Apply path requires reviewer review
+    # permission; assessment itself uses gitea.read inside the tool.
+    "cleanup_stale_review_decision_lock": {
+        "permission": "gitea.pr.review",
+        "role": "reviewer",
+    },
+    "gitea_cleanup_stale_review_decision_lock": {
+        "permission": "gitea.pr.review",
+        "role": "reviewer",
+    },
+    # #709: truthful absence-of-proof recovery (server-side auth + record + consume).
+    # Dedicated mutation capability — gitea.read is insufficient (review 434 F1).
+    "issue_irrecoverable_provenance_authorization": {
+        "permission": "gitea.decision_lock.irrecoverable_recovery",
+        "role": "reconciler",
+    },
+    "gitea_issue_irrecoverable_provenance_authorization": {
+        "permission": "gitea.decision_lock.irrecoverable_recovery",
+        "role": "reconciler",
+    },
+    "record_irrecoverable_decision_lock_provenance": {
+        "permission": "gitea.decision_lock.irrecoverable_recovery",
+        "role": "reconciler",
+    },
+    "gitea_record_irrecoverable_decision_lock_provenance": {
+        "permission": "gitea.decision_lock.irrecoverable_recovery",
+        "role": "reconciler",
+    },
+    "consume_irrecoverable_decision_lock_provenance": {
+        "permission": "gitea.pr.merge",
+        "role": "merger",
+    },
+    "gitea_consume_irrecoverable_decision_lock_provenance": {
+        "permission": "gitea.pr.merge",
+        "role": "merger",
+    },
     "delete_branch": {
         "permission": "gitea.branch.delete",
         "role": "author",
     },
+    "cleanup_merged_pr_branch": {
+        "permission": "gitea.branch.delete",
+        "role": "reconciler",
+    },
     "commit_files": {
         "permission": "gitea.repo.commit",
         "role": "author",
@@ -106,11 +171,11 @@ TASK_CAPABILITY_MAP: dict[str, dict[str, str]] = {
     },
     "reconcile_merged_cleanups": {
         "permission": "gitea.read",
-        "role": "author",
+        "role": "reconciler",
     },
     "reconciliation_cleanup": {
         "permission": "gitea.branch.delete",
-        "role": "author",
+        "role": "reconciler",
     },
     "work_issue": {
         "permission": "gitea.pr.create",
@@ -120,6 +185,103 @@ TASK_CAPABILITY_MAP: dict[str, dict[str, str]] = {
         "permission": "gitea.pr.create",
         "role": "author",
     },
+    # #600: controller-owned allocator — any authenticated profile may call;
+    # routing enforces role match to selected work. Uses control-plane DB (#613).
+    "allocate_next_work": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_allocate_next_work": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+
+    # #601 first-class lease lifecycle — inspect/list need read; mutations gate on
+    # ownership in the control-plane DB (not a separate Gitea write permission).
+    "list_workflow_leases": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_list_workflow_leases": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "inspect_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_inspect_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "adopt_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_adopt_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "release_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_release_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "expire_workflow_leases": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_expire_workflow_leases": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "abandon_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_abandon_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "reclaim_expired_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_reclaim_expired_workflow_lease": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    # #612 incident bridge — reconcile uses create_issue for apply;
+    # dry-run needs read only. Tools gate apply paths themselves.
+    "observability_reconcile_incident": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_observability_reconcile_incident": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "observability_list_projects": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_observability_list_projects": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "observability_link_issue": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+    "gitea_observability_link_issue": {
+        "permission": "gitea.read",
+        "role": "author",
+    },
+
+
     "reconcile_landed_pr": {
         "permission": "gitea.read",
         "role": "author",
@@ -142,6 +304,18 @@ TASK_CAPABILITY_MAP: dict[str, dict[str, str]] = {
         "permission": "gitea.issue.close",
         "role": "reconciler",
     },
+    "reconcile_close_superseded_pr": {
+        "permission": "gitea.pr.close",
+        "role": "reconciler",
+    },
+    "reconcile_close_satisfied_issue": {
+        "permission": "gitea.issue.close",
+        "role": "reconciler",
+    },
+    "reconcile_create_followup_issue": {
+        "permission": "gitea.issue.create",
+        "role": "reconciler",
+    },
     "post_heartbeat": {
         "permission": "gitea.issue.comment",
         "role": "author",
@@ -163,6 +337,7 @@ ISSUE_MUTATION_TOOL_TASKS: dict[str, str] = {
     "gitea_create_issue_comment": "comment_issue",
     "gitea_mark_issue": "mark_issue",
     "gitea_set_issue_labels": "set_issue_labels",
+    "gitea_create_label": "create_label",
     "gitea_commit_files": "commit_files",
 }
 

test_mcp_conn.py

@@ -1,20 +1,45 @@
 #!/usr/bin/env python3
-"""Live health-check script to verify Gitea MCP namespace connections.
+"""Offline-only MCP namespace spawn probe (NOT IDE-namespace proof).
 
-Spawns the MCP server processes as defined in the IDE's global config,
-performs the JSON-RPC handshake, and queries the tools list to verify
-that the connection is fully operational and doesn't return EOF.
+Spawns a *separate* MCP server process from config via subprocess.Popen,
+performs the JSON-RPC handshake, verifies the required tool is registered,
+and invokes that tool. Results are classified with
+``probe_source=offline_spawn``.
+
+This path is useful for offline launch/registration debugging. It does
+**not** prove the IDE-managed MCP client namespace is healthy (#543). For
+workflow gates, pass live IDE call evidence with
+``probe_source=client_namespace`` to ``gitea_assess_mcp_namespace_health``.
 """
+
+import argparse
 import json
 import os
 import subprocess
 import sys
 
-def test_connection(name, config):
+from mcp_namespace_health import REQUIRED_NAMESPACE_TOOLS, classify_namespace_probe
+
+
+def _read_json_line(proc):
+    line = proc.stdout.readline()
+    if not line:
+        stderr_content = proc.stderr.read()
+        return None, stderr_content
+    return json.loads(line), None
+
+
+def _write_message(proc, payload):
+    proc.stdin.write(json.dumps(payload) + "\n")
+    proc.stdin.flush()
+
+
+def run_connection_test(name, config, *, required_tool=None, config_path=None):
     print(f"Testing MCP connection for '{name}'...")
     command = config.get("command")
     args = config.get("args", [])
     env = config.get("env", {})
+    tool_name = required_tool or REQUIRED_NAMESPACE_TOOLS.get(name) or "gitea_whoami"
     
     # Merge current environment
     run_env = os.environ.copy()
@@ -31,8 +56,17 @@ def test_connection(name, config):
             bufsize=1,
             env=run_env
         )
-    except Exception as e:
-        print(f"  [FAIL] Failed to spawn process: {e}")
+    except Exception as exc:
+        print(f"  [FAIL] Failed to spawn process: {exc}")
+        assessment = classify_namespace_probe(
+            name,
+            required_tool=tool_name,
+            probe_result={"success": False, "error": str(exc)},
+            process={"profile": env.get("GITEA_MCP_PROFILE"), "env": env},
+            config_path=config_path,
+                probe_source="offline_spawn",
+        )
+        print(f"  diagnostics: {json.dumps(assessment['diagnostics'], sort_keys=True)}")
         return False
 
     # Send initialize request
@@ -48,26 +82,36 @@ def test_connection(name, config):
     }
     
     try:
-        proc.stdin.write(json.dumps(init_req) + "\n")
-        proc.stdin.flush()
+        _write_message(proc, init_req)
         
         # Read response
-        line = proc.stdout.readline()
-        if not line:
-            stderr_content = proc.stderr.read()
+        res, stderr_content = _read_json_line(proc)
+        if res is None:
             print(f"  [FAIL] Received EOF from process. Stderr:\n{stderr_content}")
+            assessment = classify_namespace_probe(
+                name,
+                required_tool=tool_name,
+                probe_result={"success": False, "error": stderr_content or "EOF"},
+                process={
+                    "pid": proc.pid,
+                    "profile": env.get("GITEA_MCP_PROFILE"),
+                    "env": env,
+                },
+                config_path=config_path,
+                probe_source="offline_spawn",
+            )
+            print(f"  remediation: {' '.join(assessment['remediation'])}")
             proc.terminate()
             return False
             
-        print(f"  [OK] Received initialize response: {line.strip()[:150]}...")
+        print(f"  [OK] Received initialize response: {str(res)[:150]}...")
         
         # Send initialized notification
         init_notif = {
             "jsonrpc": "2.0",
             "method": "notifications/initialized"
         }
-        proc.stdin.write(json.dumps(init_notif) + "\n")
-        proc.stdin.flush()
+        _write_message(proc, init_notif)
         
         # Send tools/list request
         list_req = {
@@ -76,16 +120,27 @@ def test_connection(name, config):
             "params": {},
             "id": 2
         }
-        proc.stdin.write(json.dumps(list_req) + "\n")
-        proc.stdin.flush()
+        _write_message(proc, list_req)
         
-        line = proc.stdout.readline()
-        if not line:
+        res, stderr_content = _read_json_line(proc)
+        if res is None:
             print("  [FAIL] Received EOF on tools/list request.")
+            assessment = classify_namespace_probe(
+                name,
+                required_tool=tool_name,
+                probe_result={"success": False, "error": stderr_content or "EOF"},
+                process={
+                    "pid": proc.pid,
+                    "profile": env.get("GITEA_MCP_PROFILE"),
+                    "env": env,
+                },
+                config_path=config_path,
+                probe_source="offline_spawn",
+            )
+            print(f"  remediation: {' '.join(assessment['remediation'])}")
             proc.terminate()
             return False
             
-        res = json.loads(line)
         if "error" in res:
             print(f"  [FAIL] Server returned error: {res['error']}")
             proc.terminate()
@@ -94,16 +149,115 @@ def test_connection(name, config):
         tools = res.get("result", {}).get("tools", [])
         tool_names = [t.get("name") for t in tools]
         print(f"  [OK] Successfully retrieved {len(tool_names)} tools: {tool_names[:5]}...")
+        if tool_name not in tool_names:
+            assessment = classify_namespace_probe(
+                name,
+                required_tool=tool_name,
+                registered_tools=tool_names,
+                probe_result={"success": False, "error": "required tool missing"},
+                process={
+                    "pid": proc.pid,
+                    "profile": env.get("GITEA_MCP_PROFILE"),
+                    "env": env,
+                },
+                config_path=config_path,
+                probe_source="offline_spawn",
+            )
+            print(f"  [FAIL] Required tool '{tool_name}' is not registered.")
+            print(f"  remediation: {' '.join(assessment['remediation'])}")
+            proc.terminate()
+            return False
+
+        call_req = {
+            "jsonrpc": "2.0",
+            "method": "tools/call",
+            "params": {"name": tool_name, "arguments": {}},
+            "id": 3,
+        }
+        _write_message(proc, call_req)
+
+        call_res, stderr_content = _read_json_line(proc)
+        if call_res is None:
+            assessment = classify_namespace_probe(
+                name,
+                required_tool=tool_name,
+                registered_tools=tool_names,
+                probe_result={"success": False, "error": stderr_content or "EOF"},
+                process={
+                    "pid": proc.pid,
+                    "profile": env.get("GITEA_MCP_PROFILE"),
+                    "env": env,
+                },
+                config_path=config_path,
+                probe_source="offline_spawn",
+            )
+            print(f"  [FAIL] Received EOF on {tool_name} invocation.")
+            print(f"  diagnostics: {json.dumps(assessment['diagnostics'], sort_keys=True)}")
+            print(f"  remediation: {' '.join(assessment['remediation'])}")
+            proc.terminate()
+            return False
+        if "error" in call_res:
+            assessment = classify_namespace_probe(
+                name,
+                required_tool=tool_name,
+                registered_tools=tool_names,
+                probe_result={"success": False, "error": call_res["error"]},
+                process={
+                    "pid": proc.pid,
+                    "profile": env.get("GITEA_MCP_PROFILE"),
+                    "env": env,
+                },
+                config_path=config_path,
+                probe_source="offline_spawn",
+            )
+            print(f"  [FAIL] {tool_name} invocation returned error: {call_res['error']}")
+            print(f"  remediation: {' '.join(assessment['remediation'])}")
+            proc.terminate()
+            return False
+
+        assessment = classify_namespace_probe(
+            name,
+            required_tool=tool_name,
+            registered_tools=tool_names,
+            probe_result={"success": True, "result": call_res.get("result")},
+            process={
+                "pid": proc.pid,
+                "profile": env.get("GITEA_MCP_PROFILE"),
+                "env": env,
+            },
+            config_path=config_path,
+                probe_source="offline_spawn",
+        )
+        print(f"  [OK] Successfully invoked required tool '{tool_name}'.")
+        print(f"  diagnostics: {json.dumps(assessment['diagnostics'], sort_keys=True)}")
         
         proc.terminate()
         return True
-    except Exception as e:
-        print(f"  [FAIL] Error during handshake: {e}")
+    except Exception as exc:
+        print(f"  [FAIL] Error during handshake: {exc}")
         proc.terminate()
         return False
 
+
 def main():
-    config_path = "/Users/jasonwalker/.gemini/config/mcp_config.json"
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--config",
+        default=os.environ.get(
+            "MCP_CONFIG_PATH",
+            os.path.expanduser("~/.gemini/config/mcp_config.json"),
+        ),
+        help="Path to MCP config JSON.",
+    )
+    parser.add_argument(
+        "--namespace",
+        action="append",
+        dest="namespaces",
+        help="Namespace to test. May be repeated.",
+    )
+    args = parser.parse_args()
+
+    config_path = args.config
     try:
         with open(config_path) as f:
             mcp_config = json.load(f)
@@ -112,10 +266,16 @@ def main():
         sys.exit(1)
         
     servers = mcp_config.get("mcpServers", {})
+    namespaces = args.namespaces or list(REQUIRED_NAMESPACE_TOOLS)
     failed = False
-    for name in ["gitea-author", "gitea-reviewer"]:
+    for name in namespaces:
         if name in servers:
-            if not test_connection(name, servers[name]):
+            if not run_connection_test(
+                name,
+                servers[name],
+                required_tool=REQUIRED_NAMESPACE_TOOLS.get(name),
+                config_path=config_path,
+            ):
                 failed = True
         else:
             print(f"Server '{name}' not found in mcp_config.json")

tests/conftest.py

@@ -1,4 +1,5 @@
 """Shared pytest fixtures for the Gitea-Tools test suite."""
+import os
 import sys
 
 import pytest
@@ -16,13 +17,66 @@ def _reset_mutation_authority(monkeypatch):
     the next test. It must never replace verify_mutation_authority with a
     no-op: individual tests that need a specific authority state set it up
     explicitly.
+
+    Session-state isolation (#559 / #590 / #594): many tests use
+    ``patch.dict(os.environ, ..., clear=True)``, which drops
+    ``GITEA_MCP_SESSION_STATE_DIR``. Relying only on the env var therefore
+    re-exposes the operator host cache
+    (``~/.cache/gitea-tools/session-state``) and its review-mutation ledger.
+    Pin ``default_state_dir`` / ``DEFAULT_STATE_DIR`` to a per-test temp dir
+    so durable load/save never touches host state even after env clears.
     """
-    monkeypatch.delenv("GITEA_SESSION_PROFILE_LOCK", raising=False)
+    for env_key in [
+        "GITEA_SESSION_PROFILE_LOCK",
+        "GITEA_ACTIVE_WORKTREE",
+        "GITEA_AUTHOR_WORKTREE",
+        "GITEA_REVIEWER_WORKTREE",
+        "GITEA_MERGER_WORKTREE",
+        "GITEA_RECONCILER_WORKTREE",
+    ]:
+        monkeypatch.delenv(env_key, raising=False)
+
     # Isolate durable session-state files so tests never share host cache (#559).
     import tempfile
 
     _state_tmp = tempfile.TemporaryDirectory(prefix="gitea-session-state-")
-    monkeypatch.setenv("GITEA_MCP_SESSION_STATE_DIR", _state_tmp.name)
+    state_dir = _state_tmp.name
+    monkeypatch.setenv("GITEA_MCP_SESSION_STATE_DIR", state_dir)
+    try:
+        import mcp_session_state
+    except Exception:
+        mcp_session_state = None
+    if mcp_session_state is not None:
+        # Survive patch.dict(..., clear=True) that drops the env var (#590).
+        # Still honor an explicit GITEA_MCP_SESSION_STATE_DIR when tests set
+        # their own temp dir (see tests/test_mcp_session_state.py).
+        monkeypatch.setattr(
+            mcp_session_state, "DEFAULT_STATE_DIR", state_dir, raising=False
+        )
+
+        def _isolated_default_state_dir(
+            _fallback: str = state_dir,
+            _env_key: str = mcp_session_state.STATE_DIR_ENV,
+        ) -> str:
+            # #695 AC2: when production native transport has pinned a session
+            # state root, that pin is authoritative even under test isolation
+            # (PR #701 redirected-state regression).
+            try:
+                import mcp_daemon_guard
+
+                pinned = mcp_daemon_guard.pinned_session_state_dir()
+                if pinned:
+                    return pinned
+            except Exception:
+                pass
+            raw = (os.environ.get(_env_key) or "").strip()
+            return raw or _fallback
+
+        monkeypatch.setattr(
+            mcp_session_state,
+            "default_state_dir",
+            _isolated_default_state_dir,
+        )
     try:
         import mcp_server
     except Exception:
@@ -32,6 +86,23 @@ def _reset_mutation_authority(monkeypatch):
     monkeypatch.setattr(mcp_server, "_MUTATION_AUTHORITY", None)
     monkeypatch.setattr(mcp_server, "_IDENTITY_CACHE", {})
     monkeypatch.setattr(mcp_server, "_REVIEW_DECISION_LOCK", None)
+    monkeypatch.setattr(mcp_server, "_LIVE_NAMESPACE_HEALTH", {})
+    monkeypatch.setattr(mcp_server, "_preflight_whoami_called", False)
+    monkeypatch.setattr(mcp_server, "_preflight_capability_called", False)
+    monkeypatch.setattr(mcp_server, "_preflight_resolved_role", None)
+    monkeypatch.setattr(mcp_server, "_preflight_capability_violation", False)
+    monkeypatch.setattr(mcp_server, "_preflight_capability_violation_files", [])
+    monkeypatch.setattr(mcp_server, "_preflight_capability_baseline_porcelain", None)
+    monkeypatch.setattr(mcp_server, "_preflight_resolved_task", None)
+    monkeypatch.setattr(mcp_server, "_preflight_reviewer_violation_files", [])
+    # Unit tests must not depend on live host MCP process health. Tests that
+    # use patch.dict(..., clear=True) drop PYTEST_CURRENT_TEST, which would
+    # otherwise re-enable the stale-runtime probe against operator daemons.
+    monkeypatch.setattr(
+        mcp_server,
+        "_check_mcp_runtimes_diagnostics",
+        lambda *args, **kwargs: [],
+    )
     try:
         import review_workflow_load
         review_workflow_load._REVIEW_WORKFLOW_LOAD = None

tests/test_allocator_service.py

@@ -0,0 +1,366 @@
+"""Tests for controller-owned allocator (#600) on control-plane DB (#613)."""
+
+from __future__ import annotations
+
+import os
+import tempfile
+import threading
+import unittest
+from concurrent.futures import ThreadPoolExecutor, as_completed
+
+from allocator_service import (
+    OUTCOME_ASSIGNED,
+    OUTCOME_BLOCKED_TERMINAL,
+    OUTCOME_NO_SAFE,
+    OUTCOME_PREVIEW,
+    OUTCOME_WAIT,
+    WorkCandidate,
+    allocate_next_work,
+    candidate_from_dict,
+    classify_skip,
+    expected_role_for_candidate,
+)
+from control_plane_db import ControlPlaneDB, InvalidWorkKindError
+
+
+class AllocatorServiceTest(unittest.TestCase):
+    def setUp(self) -> None:
+        self._tmp = tempfile.TemporaryDirectory()
+        self.db_path = os.path.join(self._tmp.name, "cp.sqlite3")
+        self.db = ControlPlaneDB(self.db_path)
+
+    def tearDown(self) -> None:
+        self._tmp.cleanup()
+
+    def _alloc(self, **kwargs):
+        defaults = dict(
+            db=self.db,
+            session_id="s-test",
+            role="author",
+            remote="prgs",
+            org="org",
+            repo="repo",
+            candidates=[],
+            apply=False,
+            profile_name="prgs-author",
+            username="jcwalker3",
+        )
+        defaults.update(kwargs)
+        return allocate_next_work(**defaults)
+
+    def test_selects_ready_issue_for_author(self) -> None:
+        cands = [
+            WorkCandidate(
+                kind="issue",
+                number=612,
+                labels=("status:ready",),
+                title="bridge",
+                dependency_unmet=True,
+                dependency_reason="downstream of #600",
+            ),
+            WorkCandidate(
+                kind="issue",
+                number=600,
+                labels=("status:ready", "type:feature"),
+                title="allocator",
+                priority=50,
+            ),
+            WorkCandidate(
+                kind="issue",
+                number=601,
+                labels=("status:blocked",),
+                title="blocked",
+                blocked=True,
+            ),
+        ]
+        res = self._alloc(candidates=cands, apply=False)
+        self.assertTrue(res["success"])
+        self.assertEqual(res["outcome"], OUTCOME_PREVIEW)
+        self.assertEqual(res["selected"]["number"], 600)
+        self.assertEqual(res["selected"]["kind"], "issue")
+        # When 600 is highest priority valid work, lower-priority blocked/dep
+        # candidates are not visited. Prove they are skipped when they sort first.
+        res2 = self._alloc(
+            candidates=[
+                WorkCandidate(
+                    kind="issue",
+                    number=612,
+                    labels=("status:ready",),
+                    priority=99,
+                    dependency_unmet=True,
+                    dependency_reason="downstream of #600",
+                ),
+                WorkCandidate(
+                    kind="issue",
+                    number=601,
+                    labels=("status:blocked",),
+                    priority=98,
+                    blocked=True,
+                ),
+                WorkCandidate(
+                    kind="issue",
+                    number=600,
+                    labels=("status:ready",),
+                    priority=1,
+                ),
+            ],
+            apply=False,
+        )
+        skipped_nums = {s["number"] for s in res2["skipped"]}
+        self.assertIn(612, skipped_nums)
+        self.assertIn(601, skipped_nums)
+        self.assertEqual(res2["selected"]["number"], 600)
+        self.assertIsNone(res["assignment"])
+        self.assertFalse(res["file_lock_only"])
+        self.assertFalse(res["comment_lease_only"])
+
+    def test_atomic_assign_and_lease_on_apply(self) -> None:
+        cands = [
+            WorkCandidate(
+                kind="issue",
+                number=600,
+                labels=("status:ready",),
+                priority=10,
+            )
+        ]
+        res = self._alloc(candidates=cands, apply=True, session_id="s-a")
+        self.assertEqual(res["outcome"], OUTCOME_ASSIGNED)
+        asn = res["assignment"]
+        self.assertEqual(asn["outcome"], "assigned")
+        self.assertIsNotNone(asn["assignment_id"])
+        self.assertIsNotNone(asn["lease_id"])
+        self.assertEqual(asn["work_number"], 600)
+        self.assertIn("implement", asn["allowed_actions"])
+        self.assertIn("merge", asn["forbidden_actions"])
+        proof = res["lease_proof"]
+        self.assertEqual(proof["source"], "control_plane_db.assign_and_lease")
+
+    def test_concurrent_allocators_no_double_assign(self) -> None:
+        cand = WorkCandidate(
+            kind="pr",
+            number=100,
+            head_sha="a" * 40,
+            priority=10,
+        )
+        # Seed work item so both race the same target
+        self.db.upsert_work_item(
+            remote="prgs",
+            org="org",
+            repo="repo",
+            kind="pr",
+            number=100,
+            current_head_sha="a" * 40,
+        )
+        results = []
+        lock = threading.Lock()
+
+        def worker(sid: str):
+            r = allocate_next_work(
+                self.db,
+                session_id=sid,
+                role="reviewer",
+                remote="prgs",
+                org="org",
+                repo="repo",
+                candidates=[cand],
+                apply=True,
+                profile_name="prgs-reviewer",
+            )
+            with lock:
+                results.append(r)
+
+        with ThreadPoolExecutor(max_workers=2) as pool:
+            futs = [pool.submit(worker, f"s-{i}") for i in range(2)]
+            for f in as_completed(futs):
+                f.result()
+
+        outcomes = [r["outcome"] for r in results]
+        self.assertEqual(outcomes.count(OUTCOME_ASSIGNED), 1, outcomes)
+        self.assertEqual(outcomes.count(OUTCOME_WAIT), 1, outcomes)
+
+    def test_blocked_and_dependency_skipped(self) -> None:
+        cands = [
+            WorkCandidate(kind="issue", number=1, blocked=True, labels=("status:blocked",)),
+            WorkCandidate(
+                kind="issue",
+                number=612,
+                labels=("status:ready",),
+                dependency_unmet=True,
+                dependency_reason="downstream of #600",
+            ),
+        ]
+        res = self._alloc(candidates=cands, apply=True, role="author")
+        self.assertEqual(res["outcome"], OUTCOME_NO_SAFE)
+        self.assertIsNone(res["selected"])
+        reasons = " ".join(s["reason"] for s in res["skipped"])
+        self.assertIn("blocked", reasons.lower())
+        self.assertIn("600", reasons)
+
+    def test_already_leased_returns_wait(self) -> None:
+        self.db.upsert_session(session_id="owner", role="author")
+        self.db.assign_and_lease(
+            session_id="owner",
+            role="author",
+            remote="prgs",
+            org="org",
+            repo="repo",
+            kind="issue",
+            number=50,
+        )
+        res = self._alloc(
+            session_id="other",
+            candidates=[
+                WorkCandidate(kind="issue", number=50, labels=("status:ready",))
+            ],
+            apply=True,
+        )
+        self.assertEqual(res["outcome"], OUTCOME_WAIT)
+        self.assertEqual(res["owner_session_id"], "owner")
+
+    def test_role_ineligible_skipped(self) -> None:
+        # PR with current-head REQUEST_CHANGES expects author, not reviewer.
+        cand = WorkCandidate(
+            kind="pr",
+            number=9,
+            head_sha="b" * 40,
+            request_changes_current_head=True,
+        )
+        self.assertEqual(expected_role_for_candidate(cand), "author")
+        res = self._alloc(
+            role="reviewer",
+            profile_name="prgs-reviewer",
+            candidates=[cand],
+            apply=False,
+        )
+        self.assertEqual(res["outcome"], OUTCOME_NO_SAFE)
+        self.assertTrue(any("expects role" in s["reason"] for s in res["skipped"]))
+
+    def test_terminal_lock_blocks_downstream_reviewer_prs(self) -> None:
+        self.db.set_terminal_lock(
+            remote="prgs",
+            org="org",
+            repo="repo",
+            terminal_pr=10,
+            decision="request_changes",
+            status="active",
+        )
+        cands = [
+            WorkCandidate(kind="pr", number=11, head_sha="c" * 40, priority=5),
+            WorkCandidate(kind="pr", number=10, head_sha="d" * 40, priority=1),
+        ]
+        res = self._alloc(
+            role="reviewer",
+            profile_name="prgs-reviewer",
+            candidates=cands,
+            apply=False,
+        )
+        # Terminal PR #10 is selectable; #11 skipped for terminal path.
+        self.assertEqual(res["selected"]["number"], 10)
+        self.assertTrue(
+            any(
+                s["number"] == 11 and "terminal-review lock" in s["reason"]
+                for s in res["skipped"]
+            )
+        )
+
+    def test_terminal_lock_blocks_all_when_only_downstream(self) -> None:
+        self.db.set_terminal_lock(
+            remote="prgs",
+            org="org",
+            repo="repo",
+            terminal_pr=10,
+            decision="request_changes",
+            status="active",
+        )
+        res = self._alloc(
+            role="reviewer",
+            profile_name="prgs-reviewer",
+            candidates=[
+                WorkCandidate(kind="pr", number=99, head_sha="e" * 40),
+            ],
+            apply=False,
+        )
+        self.assertEqual(res["outcome"], OUTCOME_BLOCKED_TERMINAL)
+
+    def test_no_work_structured(self) -> None:
+        res = self._alloc(candidates=[], apply=True)
+        self.assertEqual(res["outcome"], OUTCOME_NO_SAFE)
+        self.assertIsNone(res["selected"])
+        self.assertTrue(res["reasons"])
+
+    def test_rejects_incident_kind(self) -> None:
+        with self.assertRaises(InvalidWorkKindError):
+            WorkCandidate(kind="sentry_incident", number=1)
+
+    def test_612_downstream_marker_in_result(self) -> None:
+        res = self._alloc(
+            candidates=[
+                WorkCandidate(kind="issue", number=600, labels=("status:ready",))
+            ],
+            apply=True,
+        )
+        self.assertIn("612", res.get("downstream_note", ""))
+
+    def test_substrate_not_file_or_comment_lease(self) -> None:
+        res = self._alloc(
+            candidates=[
+                WorkCandidate(kind="issue", number=1, labels=("status:ready",))
+            ],
+            apply=True,
+        )
+        self.assertEqual(res["substrate"], "control_plane_db")
+        self.assertFalse(res["file_lock_only"])
+        self.assertFalse(res["comment_lease_only"])
+        self.assertEqual(
+            res["lease_proof"]["source"], "control_plane_db.assign_and_lease"
+        )
+
+    def test_candidate_from_dict(self) -> None:
+        c = candidate_from_dict(
+            {
+                "kind": "pr",
+                "number": 7,
+                "head_sha": "f" * 40,
+                "approval_on_current_head": True,
+                "mergeable": True,
+            }
+        )
+        self.assertEqual(expected_role_for_candidate(c), "merger")
+
+    def test_merger_gets_clean_approval(self) -> None:
+        c = WorkCandidate(
+            kind="pr",
+            number=3,
+            head_sha="1" * 40,
+            approval_on_current_head=True,
+            mergeable=True,
+            priority=100,
+        )
+        res = self._alloc(
+            role="merger",
+            profile_name="prgs-merger",
+            candidates=[c],
+            apply=True,
+            session_id="merger-1",
+        )
+        self.assertEqual(res["outcome"], OUTCOME_ASSIGNED)
+        self.assertEqual(res["assignment"]["work_number"], 3)
+        self.assertIn("merge", res["assignment"]["allowed_actions"])
+
+    def test_db_unavailable_fails_closed(self) -> None:
+        res = allocate_next_work(
+            None,  # type: ignore[arg-type]
+            session_id="x",
+            role="author",
+            remote="prgs",
+            org="o",
+            repo="r",
+            candidates=[],
+            apply=True,
+        )
+        self.assertFalse(res["success"])
+        self.assertIn("unavailable", res["reasons"][0].lower())
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_api_reliability.py

@@ -79,41 +79,46 @@ class TestApiRequestFailures(unittest.TestCase):
     @patch("gitea_auth.urllib.request.urlopen")
     def test_timeout_converted_to_runtimeerror(self, mock_open):
         mock_open.side_effect = TimeoutError("timed out")
-        with self.assertRaises(RuntimeError) as ctx:
+        with self.assertRaises(gitea_auth.GiteaNetworkError) as ctx:
             gitea_auth.api_request("GET", URL, FAKE_AUTH)
-        self.assertIn("network error contacting Gitea", str(ctx.exception))
+        # Fixed message only (#699) — still a RuntimeError subclass.
+        self.assertIsInstance(ctx.exception, RuntimeError)
+        self.assertEqual(str(ctx.exception), "Network error contacting Gitea")
 
     @patch("gitea_auth.urllib.request.urlopen")
     def test_dns_network_failure_converted(self, mock_open):
         mock_open.side_effect = urllib.error.URLError("Name or service not known")
-        with self.assertRaises(RuntimeError) as ctx:
+        with self.assertRaises(gitea_auth.GiteaNetworkError) as ctx:
             gitea_auth.api_request("GET", URL, FAKE_AUTH)
-        self.assertIn("network error contacting Gitea", str(ctx.exception))
+        self.assertEqual(str(ctx.exception), "Network error contacting Gitea")
 
     @patch("gitea_auth.urllib.request.urlopen")
     def test_502_upstream_message(self, mock_open):
         mock_open.side_effect = http_error(502, "bad gateway")
-        with self.assertRaises(RuntimeError) as ctx:
+        with self.assertRaises(gitea_auth.GiteaHttpError) as ctx:
             gitea_auth.api_request("GET", URL, FAKE_AUTH)
         msg = str(ctx.exception)
-        self.assertIn("HTTP 502", msg)
-        self.assertIn("upstream unavailable", msg)
+        self.assertEqual(msg, "Gitea upstream unavailable")
+        self.assertEqual(ctx.exception.reason_code, "upstream_unavailable")
+        self.assertEqual(ctx.exception.http_status, 502)
 
     @patch("gitea_auth.urllib.request.urlopen")
     def test_503_upstream_message(self, mock_open):
         mock_open.side_effect = http_error(503, "")
-        with self.assertRaises(RuntimeError) as ctx:
+        with self.assertRaises(gitea_auth.GiteaHttpError) as ctx:
             gitea_auth.api_request("GET", URL, FAKE_AUTH)
-        self.assertIn("HTTP 503", str(ctx.exception))
-        self.assertIn("upstream unavailable", str(ctx.exception))
+        self.assertEqual(str(ctx.exception), "Gitea upstream unavailable")
+        self.assertEqual(ctx.exception.http_status, 503)
 
     @patch("gitea_auth.urllib.request.urlopen")
     def test_malformed_error_payload_does_not_crash(self, mock_open):
-        # Non-JSON garbage error body must still yield a clean RuntimeError.
+        # Non-JSON garbage error body must still yield a clean typed error
+        # with a fixed message (no body echo — #699).
         mock_open.side_effect = http_error(500, "<html>garbage</html>")
-        with self.assertRaises(RuntimeError) as ctx:
+        with self.assertRaises(gitea_auth.GiteaHttpError) as ctx:
             gitea_auth.api_request("GET", URL, FAKE_AUTH)
-        self.assertIn("HTTP 500", str(ctx.exception))
+        self.assertEqual(str(ctx.exception), "Gitea HTTP request failed")
+        self.assertNotIn("<html>", str(ctx.exception))
 
     @patch("gitea_auth.urllib.request.urlopen")
     def test_malformed_success_json_raises_clean_error(self, mock_open):
@@ -126,16 +131,17 @@ class TestApiRequestFailures(unittest.TestCase):
     def test_no_secret_leak_in_error_body(self, mock_open):
         mock_open.side_effect = http_error(
             400, "failed: token supersecret123 rejected")
-        with self.assertRaises(RuntimeError) as ctx:
+        with self.assertRaises(gitea_auth.GiteaHttpError) as ctx:
             gitea_auth.api_request("GET", URL, FAKE_AUTH)
         msg = str(ctx.exception)
         self.assertNotIn("supersecret123", msg)
-        self.assertIn(gitea_audit.REDACTED, msg)
+        # Fixed message — body never appears (stronger than redaction).
+        self.assertEqual(msg, "Gitea HTTP request failed")
 
     @patch("gitea_auth.urllib.request.urlopen")
     def test_auth_header_never_in_error(self, mock_open):
         mock_open.side_effect = http_error(400, "bad request")
-        with self.assertRaises(RuntimeError) as ctx:
+        with self.assertRaises(gitea_auth.GiteaHttpError) as ctx:
             gitea_auth.api_request("GET", URL, FAKE_AUTH)
         self.assertNotIn(FAKE_AUTH, str(ctx.exception))
 

tests/test_audit.py

@@ -329,13 +329,17 @@ class TestGatedToolAudit(_AuditWiringBase):
     @patch("mcp_server.api_request")
     @patch("mcp_server.get_auth_header", return_value=FAKE_AUTH)
     def test_merge_success_audited(self, _auth, mock_api):
-        # user, pr, feedback pr+reviews, merge POST, readback.
+        # user, pr, eligibility feedback pr+reviews (#695), gate-7 feedback
+        # pr+reviews, merge POST, readback.
+        approval = [{
+            "id": 1, "user": {"login": "reviewer-bot"}, "state": "APPROVED",
+            "commit_id": "abc123", "submitted_at": "2026-07-06T10:00:00Z",
+            "dismissed": False,
+        }]
         mock_api.side_effect = [
             {"login": "merger-bot"}, self._pr("author-bot"),
-            self._pr("author-bot"),
-            [{"id": 1, "user": {"login": "reviewer-bot"}, "state": "APPROVED",
-              "commit_id": "abc123", "submitted_at": "2026-07-06T10:00:00Z",
-              "dismissed": False}],
+            self._pr("author-bot"), approval,  # eligibility merge feedback
+            self._pr("author-bot"), approval,  # gate 7 feedback
             {}, {"merged_commit_sha": "c1"},
         ]
         env = self._env(GITEA_PROFILE_NAME="gitea-merger",

tests/test_audit_reconciliation_mode.py

@@ -27,7 +27,13 @@ from task_capability_map import required_permission, required_role
 
 DELETE_PROFILE = {
     "profile_name": "prgs-author-delete",
-    "allowed_operations": ["gitea.read", "gitea.branch.delete"],
+    "role": "author",
+    "allowed_operations": [
+        "gitea.read",
+        "gitea.pr.create",
+        "gitea.branch.push",
+        "gitea.branch.delete",
+    ],
     "forbidden_operations": [],
     "audit_label": "prgs-author-delete",
 }
@@ -284,7 +290,7 @@ class TestTaskCapabilityMap(unittest.TestCase):
             required_permission("reconciliation_cleanup"),
             "gitea.branch.delete",
         )
-        self.assertEqual(required_role("reconciliation_cleanup"), "author")
+        self.assertEqual(required_role("reconciliation_cleanup"), "reconciler")
 
 
 if __name__ == "__main__":

tests/test_author_mutation_worktree.py

@@ -81,13 +81,15 @@ class TestPreflightIntegration(unittest.TestCase):
         mcp_server._preflight_resolved_role = "author"
         control_root = "/repo/Gitea-Tools"
         with mock.patch.object(mcp_server, "PROJECT_ROOT", control_root):
-            with mock.patch.dict(
-                "os.environ",
-                {"GITEA_TEST_PORCELAIN": ""},
-                clear=False,
-            ):
-                with self.assertRaises(RuntimeError) as ctx:
-                    mcp_server.verify_preflight_purity()
+            with mock.patch.object(mcp_server, "_enforce_root_checkout_guard"):
+                with mock.patch("gitea_auth.get_profile", return_value={"profile_name": "gitea-author"}):
+                    with mock.patch.dict(
+                        "os.environ",
+                        {"GITEA_TEST_PORCELAIN": ""},
+                        clear=False,
+                    ):
+                        with self.assertRaises(RuntimeError) as ctx:
+                            mcp_server.verify_preflight_purity()
         self.assertIn("Branches-only mutation guard", str(ctx.exception))
 
     def test_verify_preflight_allows_branches_worktree(self):
@@ -97,12 +99,13 @@ class TestPreflightIntegration(unittest.TestCase):
         mcp_server._preflight_capability_called = True
         mcp_server._preflight_resolved_role = "author"
         worktree = "/repo/Gitea-Tools/branches/issue-274"
-        with mock.patch.dict(
-            "os.environ",
-            {"GITEA_TEST_PORCELAIN": ""},
-            clear=False,
-        ):
-            mcp_server.verify_preflight_purity(worktree_path=worktree)
+        with mock.patch.object(mcp_server, "_enforce_root_checkout_guard"):
+            with mock.patch.dict(
+                "os.environ",
+                {"GITEA_TEST_PORCELAIN": ""},
+                clear=False,
+            ):
+                mcp_server.verify_preflight_purity(worktree_path=worktree)
 
 
 if __name__ == "__main__":