feat: gate author work on early duplicate-work detection (Closes #400)

Add author_duplicate_work_gate and enforce it at claim, lock, and PR
creation. Expose gitea_assess_author_duplicate_work for pre-commit/push
checks and extend work-issue final-report verification.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

author_duplicate_work_gate.py

@@ -0,0 +1,198 @@
+"""Early duplicate-work detection for author work-issue flows (#400)."""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+from issue_claim_heartbeat import (
+    _linked_open_pr,
+    _matching_branch_names,
+    classify_issue_claim,
+)
+
+STAGES = (
+    "claim",
+    "lock",
+    "worktree",
+    "edit",
+    "commit",
+    "push",
+    "create_pr",
+)
+
+ELIGIBILITY_OPEN_PR_EXISTS = "OPEN_PR_EXISTS"
+ELIGIBILITY_DUPLICATE_BRANCH_EXISTS = "DUPLICATE_BRANCH_EXISTS"
+ELIGIBILITY_ACTIVE_CLAIM = "ACTIVE_CLAIM_BY_OTHER"
+ELIGIBILITY_CLEAR = "CLEAR"
+
+
+def assess_author_duplicate_work(
+    issue_number: int,
+    *,
+    stage: str,
+    open_prs: list[dict] | None = None,
+    branch_names: list[str] | None = None,
+    claim_entry: dict | None = None,
+    matching_branches: list[str] | None = None,
+    allow_stale_takeover: bool = False,
+) -> dict[str, Any]:
+    """Fail closed when duplicate work is detected before author mutations (#400)."""
+    stage_norm = (stage or "").strip().lower()
+    if stage_norm not in STAGES:
+        return {
+            "allowed": False,
+            "block": True,
+            "eligibility_class": "INVALID_STAGE",
+            "stage": stage_norm or None,
+            "reasons": [f"unknown duplicate-work stage {stage!r}"],
+            "safe_next_action": f"use one of: {', '.join(STAGES)}",
+        }
+
+    prs = list(open_prs or [])
+    linked_pr = _linked_open_pr(int(issue_number), prs)
+    branches = list(
+        matching_branches
+        if matching_branches is not None
+        else _matching_branch_names(int(issue_number), list(branch_names or []))
+    )
+
+    reasons: list[str] = []
+    eligibility = ELIGIBILITY_CLEAR
+
+    if linked_pr:
+        eligibility = ELIGIBILITY_OPEN_PR_EXISTS
+        reasons.append(
+            f"open PR #{linked_pr.get('number')} already covers issue "
+            f"#{issue_number}"
+        )
+
+    if branches and stage_norm in {"claim", "lock", "worktree", "edit"}:
+        if eligibility == ELIGIBILITY_CLEAR:
+            eligibility = ELIGIBILITY_DUPLICATE_BRANCH_EXISTS
+        reasons.append(
+            f"remote branch(es) already exist for issue #{issue_number}: "
+            f"{', '.join(branches)}"
+        )
+
+    entry = claim_entry or {}
+    claim_status = (entry.get("status") or "").strip()
+    if claim_status in {"active", "awaiting_review"} and stage_norm == "claim":
+        if entry.get("reclaimable") and allow_stale_takeover:
+            pass
+        elif claim_status == "active" and not entry.get("reclaimable"):
+            if eligibility == ELIGIBILITY_CLEAR:
+                eligibility = ELIGIBILITY_ACTIVE_CLAIM
+            reasons.append(
+                f"issue #{issue_number} has active claim "
+                f"(status={claim_status})"
+            )
+    elif claim_status == "awaiting_review" and stage_norm == "claim":
+        if not linked_pr:
+            reasons.append(
+                f"issue #{issue_number} is awaiting_review but no linked open PR "
+                "was supplied for duplicate-work proof"
+            )
+
+    allowed = not reasons
+    outcome = "duplicate_work_prevented" if not allowed else "clear"
+    if stage_norm == "create_pr" and not allowed:
+        outcome = "duplicate_pr_prevented"
+    elif stage_norm in {"commit", "push"} and not allowed:
+        outcome = "duplicate_push_prevented" if stage_norm == "push" else "duplicate_commit_prevented"
+
+    return {
+        "allowed": allowed,
+        "block": not allowed,
+        "eligibility_class": eligibility if not allowed else ELIGIBILITY_CLEAR,
+        "stage": stage_norm,
+        "linked_open_pr": linked_pr.get("number") if linked_pr else None,
+        "matching_branches": branches,
+        "claim_status": claim_status or None,
+        "outcome": outcome,
+        "reasons": reasons,
+        "safe_next_action": (
+            "stop without edits/commit/push/PR; produce reconciliation handoff "
+            "preserving local work only"
+            if not allowed
+            else "proceed"
+        ),
+    }
+
+
+def build_claim_entry_from_classification(classification: dict) -> dict:
+    """Map ``classify_issue_claim`` output to gate claim metadata."""
+    return {
+        "status": classification.get("status"),
+        "reclaimable": classification.get("reclaimable"),
+        "linked_open_pr": classification.get("linked_open_pr"),
+    }
+
+
+_DUPLICATE_OUTCOME_RE = re.compile(
+    r"(duplicate\s+(?:pr|branch|commit|push)\s+prevented|"
+    r"duplicate\s+work\s+not\s+prevented|reconciliation\s+handoff)",
+    re.IGNORECASE,
+)
+
+
+def assess_work_issue_duplicate_prevention_report(report_text: str) -> dict:
+    """#400: work-issue reports must state duplicate-work prevention outcome."""
+    text = report_text or ""
+    if "duplicate work" not in text.lower() and "duplicate pr" not in text.lower():
+        return {
+            "proven": True,
+            "block": False,
+            "reasons": [],
+            "safe_next_action": "proceed",
+        }
+    if _DUPLICATE_OUTCOME_RE.search(text):
+        return {
+            "proven": True,
+            "block": False,
+            "reasons": [],
+            "safe_next_action": "proceed",
+        }
+    return {
+        "proven": False,
+        "block": True,
+        "reasons": [
+            "duplicate-work discussion must name a prevention outcome "
+            "(duplicate PR/branch/commit/push prevented, or duplicate work "
+            "not prevented, or reconciliation handoff)"
+        ],
+        "safe_next_action": "state exact duplicate-work prevention class in final report",
+    }
+
+
+def classify_and_assess(
+    issue: dict,
+    *,
+    stage: str,
+    comments: list[dict] | None = None,
+    open_prs: list[dict] | None = None,
+    branch_names: list[str] | None = None,
+    allow_stale_takeover: bool = False,
+) -> dict[str, Any]:
+    """Combine claim classification with duplicate-work gate assessment."""
+    issue_number = int(issue.get("number") or 0)
+    claim = classify_issue_claim(
+        issue=issue,
+        comments=comments or [],
+        open_prs=open_prs or [],
+        branch_names=branch_names or [],
+    )
+    assessment = assess_author_duplicate_work(
+        issue_number,
+        stage=stage,
+        open_prs=open_prs,
+        branch_names=branch_names,
+        claim_entry=build_claim_entry_from_classification(claim),
+        matching_branches=claim.get("matching_branches"),
+        allow_stale_takeover=allow_stale_takeover,
+    )
+    return {
+        "issue_number": issue_number,
+        "claim": claim,
+        "duplicate_work": assessment,
+    }

final_report_validator.py

@@ -456,6 +456,40 @@ def _rule_reviewer_git_fetch_readonly(
     return []
 
 
+def _rule_reviewer_validation_failure_history(
+    report_text: str,
+    *,
+    validation_session: dict | None = None,
+) -> list[dict[str, str]]:
+    from reviewer_validation_failure_history import (
+        assess_validation_failure_history_report,
+    )
+
+    session = validation_session or {}
+    observed = session.get("observed_failures") or []
+    if not observed:
+        return []
+
+    result = assess_validation_failure_history_report(
+        report_text,
+        validation_session=session,
+    )
+    if result.get("proven"):
+        return []
+    severity = "block" if result.get("violations") else "downgrade"
+    return [
+        validator_finding(
+            "reviewer.validation_failure_history",
+            severity,
+            "Validation failure history",
+            reason,
+            result.get("safe_next_action")
+            or "document every observed validation failure before claiming pass",
+        )
+        for reason in (result.get("violations") or result.get("reasons") or ["incomplete"])
+    ]
+
+
 def _rule_reviewer_validation_command(report_text: str) -> list[dict[str, str]]:
     text = report_text or ""
     if not _BARE_PYTEST_RE.search(text):
@@ -864,6 +898,7 @@ _RULES_BY_TASK: dict[str, list[Callable[..., list[dict[str, str]]]]] = {
         _rule_reviewer_mutation_categories,
         _rule_reviewer_git_fetch_readonly,
         _rule_reviewer_validation_command,
+        _rule_reviewer_validation_failure_history,
         _rule_reviewer_validation_structured,
         _rule_reviewer_linked_issue,
         _rule_reviewer_baseline_on_failure,
@@ -970,6 +1005,7 @@ def assess_final_report_validator(
     local_edits: bool = False,
     issue_filing_lock: dict | None = None,
     session_pr_opened: bool = False,
+    validation_session: dict | None = None,
 ) -> dict[str, Any]:
     """Validate final-report text against task-specific proof rules (#327).
 
@@ -1025,6 +1061,7 @@ def assess_final_report_validator(
         "mutations_observed": mutations_observed,
         "local_edits": local_edits,
         "session_pr_opened": session_pr_opened,
+        "validation_session": validation_session,
     }
 
     for rule in _RULES_BY_TASK.get(normalized_kind, ()):

gitea_mcp_server.py

@@ -503,6 +503,7 @@ import issue_lock_worktree  # noqa: E402
 import already_landed_reconcile  # noqa: E402
 import author_mutation_worktree  # noqa: E402
 import issue_claim_heartbeat  # noqa: E402
+import author_duplicate_work_gate  # noqa: E402
 import merged_cleanup_reconcile  # noqa: E402
 import reconciler_profile  # noqa: E402
 import reconciliation_workflow  # noqa: E402
@@ -1049,6 +1050,76 @@ def gitea_create_issue(
     return _with_optional_url({"number": data["number"]}, data.get("html_url"))
 
 
+def _list_repo_branch_names(h: str, o: str, r: str, auth: str, *, limit: int = 200) -> list[str]:
+    branches = api_get_all(f"{repo_api_url(h, o, r)}/branches", auth, limit=limit)
+    return [_branch_entry_name(branch) for branch in branches]
+
+
+def _gather_author_duplicate_work_context(
+    issue_number: int,
+    *,
+    h: str,
+    o: str,
+    r: str,
+    auth: str,
+    exclude_branch_name: str | None = None,
+) -> dict:
+    base = repo_api_url(h, o, r)
+    issue = api_request("GET", f"{base}/issues/{issue_number}", auth)
+    comments = api_request("GET", f"{base}/issues/{issue_number}/comments", auth) or []
+    open_prs = api_get_all(f"{base}/pulls?state=open", auth)
+    branch_names = _list_repo_branch_names(h, o, r, auth)
+    if exclude_branch_name:
+        branch_names = [
+            name for name in branch_names
+            if name != exclude_branch_name
+        ]
+    return {
+        "issue": issue,
+        "comments": comments,
+        "open_prs": open_prs,
+        "branch_names": branch_names,
+    }
+
+
+def _enforce_author_duplicate_work_gate(
+    issue_number: int,
+    stage: str,
+    *,
+    h: str,
+    o: str,
+    r: str,
+    auth: str,
+    allow_stale_takeover: bool = False,
+    exclude_branch_name: str | None = None,
+) -> dict:
+    """Fail closed when duplicate work is detected (#400)."""
+    ctx = _gather_author_duplicate_work_context(
+        issue_number,
+        h=h,
+        o=o,
+        r=r,
+        auth=auth,
+        exclude_branch_name=exclude_branch_name,
+    )
+    result = author_duplicate_work_gate.classify_and_assess(
+        ctx["issue"],
+        stage=stage,
+        comments=ctx["comments"],
+        open_prs=ctx["open_prs"],
+        branch_names=ctx["branch_names"],
+        allow_stale_takeover=allow_stale_takeover,
+    )
+    assessment = result.get("duplicate_work") or {}
+    if assessment.get("block"):
+        reasons = "; ".join(assessment.get("reasons") or ["duplicate work detected"])
+        raise RuntimeError(
+            f"Author duplicate-work gate (#400) blocked at stage '{stage}': "
+            f"{reasons} (fail closed)"
+        )
+    return result
+
+
 @mcp.tool()
 def gitea_lock_issue(
     issue_number: int,
@@ -1111,48 +1182,17 @@ def gitea_lock_issue(
             issue_lock_worktree.format_issue_lock_worktree_error(lock_assessment)
         )
 
-    # 2. Check if the issue already has an open PR (reuse protection)
     h, o, r = _resolve(remote, host, org, repo)
     auth = _auth(h)
-    url = f"{repo_api_url(h, o, r)}/pulls?state=open"
-
-    try:
-        prs = api_get_all(url, auth)
-    except Exception as e:
-        raise RuntimeError(f"Could not list open PRs to verify issue lock: {e}")
-
-    for pr in prs:
-        pr_head = pr.get("head", {}).get("ref", "")
-        pr_title = pr.get("title", "")
-        pr_body = pr.get("body", "")
-
-        if expected_pattern in pr_head:
-            raise ValueError(
-                f"Issue #{issue_number} is already tied to an open PR (PR #{pr.get('number')}, branch '{pr_head}') (fail closed)"
-            )
-
-        patterns = [
-            f"closes #{issue_number}",
-            f"fixes #{issue_number}",
-        ]
-        text_to_check = f"{pr_title} {pr_body}".lower()
-        if any(p in text_to_check for p in patterns):
-            raise ValueError(
-                f"Issue #{issue_number} is already tied to an open PR (PR #{pr.get('number')}) via Closes/Fixes reference (fail closed)"
-            )
-
-    branch_url = f"{repo_api_url(h, o, r)}/branches"
-    try:
-        branches = api_get_all(branch_url, auth)
-    except Exception as e:
-        raise RuntimeError(f"Could not list branches to verify issue lock: {e}")
-    for branch in branches:
-        name = _branch_entry_name(branch)
-        if expected_pattern in name:
-            raise ValueError(
-                f"Issue #{issue_number} already has matching branch '{name}' "
-                "(fail closed)"
-            )
+    _enforce_author_duplicate_work_gate(
+        issue_number,
+        "lock",
+        h=h,
+        o=o,
+        r=r,
+        auth=auth,
+        exclude_branch_name=branch_name,
+    )
 
     work_lease = _build_author_issue_work_lease(
         issue_number=issue_number,
@@ -1245,7 +1285,7 @@ def gitea_create_pr(
     )
     if blocked:
         return blocked
-    verify_preflight_purity(remote)
+    verify_preflight_purity(remote, worktree_path=worktree_path)
     h, o, r = _resolve(remote, host, org, repo)
 
     # ── Issue Lock Validation (Issue #194 / #196) ──
@@ -1273,6 +1313,17 @@ def gitea_create_pr(
             f"PR head branch '{head}' does not match locked branch '{locked_branch}' (fail closed)"
         )
 
+    auth = _auth(h)
+    _enforce_author_duplicate_work_gate(
+        int(locked_issue),
+        "create_pr",
+        h=h,
+        o=o,
+        r=r,
+        auth=auth,
+        exclude_branch_name=locked_branch,
+    )
+
     # Check for forbidden terms anywhere in title/body
     forbidden_terms = ["equivalent", "related", "same as"]
     text_to_check = f"{title} {body}".lower()
@@ -1289,7 +1340,6 @@ def gitea_create_pr(
             f"PR title or body must contain 'Closes #{locked_issue}' or 'Fixes #{locked_issue}' exactly to ensure durable tracking (fail closed)"
         )
 
-    auth = _auth(h)
     url = f"{repo_api_url(h, o, r)}/pulls"
     payload = {"title": title, "body": body, "head": head, "base": base}
     meta = {"title": title, "head": head, "base": base}
@@ -3464,16 +3514,32 @@ def gitea_delete_branch(
         repo: Override the repository name.
 
     Returns:
-        dict with 'success' and 'message'.
+        dict with 'success' and 'message'; on a permission block,
+        'success'/'performed' False, 'reasons', and a structured
+        'permission_report' (#142) with no API call made.
     """
+    gate_reasons = _profile_operation_gate("gitea.branch.delete")
+    if gate_reasons:
+        return {
+            "success": False,
+            "performed": False,
+            "required_permission": "gitea.branch.delete",
+            "reasons": gate_reasons,
+            "permission_report": _permission_block_report("gitea.branch.delete"),
+        }
+
     verify_preflight_purity(remote)
     h, o, r = _resolve(remote, host, org, repo)
     auth = _auth(h)
     import urllib.parse
     encoded_branch = urllib.parse.quote(branch, safe="")
     url = f"{repo_api_url(h, o, r)}/branches/{encoded_branch}"
+    request_metadata = {
+        "branch": branch,
+        "required_permission": "gitea.branch.delete",
+    }
     with _audited("delete_branch", host=h, remote=remote, org=o, repo=r,
-                  target_branch=branch, request_metadata={"branch": branch}):
+                  target_branch=branch, request_metadata=request_metadata):
         api_request("DELETE", url, auth)
     return {"success": True, "message": f"Remote branch '{branch}' deleted."}
 
@@ -5510,6 +5576,7 @@ def gitea_validate_review_final_report(
     action_log: list[dict] | None = None,
     mutations_observed: bool = False,
     local_edits: bool = False,
+    validation_session: dict | None = None,
 ) -> dict:
     """Read-only: machine-validate reviewer final report schema before output (#391).
 
@@ -5527,6 +5594,7 @@ def gitea_validate_review_final_report(
         action_log=action_log,
         mutations_observed=mutations_observed,
         local_edits=local_edits,
+        validation_session=validation_session,
     )
 
 
@@ -5985,6 +6053,14 @@ def gitea_mark_issue(
         )
 
     if action == "start":
+        _enforce_author_duplicate_work_gate(
+            issue_number,
+            "claim",
+            h=h,
+            o=o,
+            r=r,
+            auth=auth,
+        )
         with _audited("label_issue", host=h, remote=remote, org=o, repo=r,
                       issue_number=issue_number,
                       request_metadata={"op": "add", "label": "status:in-progress"}):
@@ -6066,6 +6142,65 @@ def gitea_post_heartbeat(
     )
 
 
+@mcp.tool()
+def gitea_assess_author_duplicate_work(
+    issue_number: int,
+    stage: str,
+    remote: str = "dadeschools",
+    host: str | None = None,
+    org: str | None = None,
+    repo: str | None = None,
+    allow_stale_takeover: bool = False,
+    exclude_branch_name: str | None = None,
+) -> dict:
+    """Read-only: assess duplicate-work risk before author mutations (#400).
+
+    Call at claim, lock, worktree, edit, commit, push, and create_pr stages.
+    """
+    read_block = _profile_operation_gate("gitea.read")
+    if read_block:
+        return {
+            "success": False,
+            "reasons": read_block,
+            "permission_report": _permission_block_report("gitea.read"),
+        }
+    h, o, r = _resolve(remote, host, org, repo)
+    auth = _auth(h)
+    ctx = _gather_author_duplicate_work_context(
+        issue_number,
+        h=h,
+        o=o,
+        r=r,
+        auth=auth,
+        exclude_branch_name=exclude_branch_name,
+    )
+    result = author_duplicate_work_gate.classify_and_assess(
+        ctx["issue"],
+        stage=stage,
+        comments=ctx["comments"],
+        open_prs=ctx["open_prs"],
+        branch_names=ctx["branch_names"],
+        allow_stale_takeover=allow_stale_takeover,
+    )
+    assessment = result.get("duplicate_work") or {}
+    return {
+        "success": True,
+        "performed": False,
+        "issue_number": issue_number,
+        "stage": stage,
+        "allowed": assessment.get("allowed"),
+        "block": assessment.get("block"),
+        "eligibility_class": assessment.get("eligibility_class"),
+        "outcome": assessment.get("outcome"),
+        "linked_open_pr": assessment.get("linked_open_pr"),
+        "matching_branches": assessment.get("matching_branches"),
+        "claim_status": assessment.get("claim_status"),
+        "reasons": assessment.get("reasons"),
+        "safe_next_action": assessment.get("safe_next_action"),
+        "claim": result.get("claim"),
+    }
+
+
 @mcp.tool()
 def gitea_reconcile_issue_claims(
     state: str = "open",

review_final_report_schema.py

@@ -301,6 +301,7 @@ def assess_review_final_report_schema(
     action_log: list[dict] | None = None,
     mutations_observed: bool = False,
     local_edits: bool = False,
+    validation_session: dict | None = None,
 ) -> dict[str, Any]:
     """Validate reviewer final report text before session completion (#391)."""
     base = assess_final_report_validator(
@@ -312,6 +313,7 @@ def assess_review_final_report_schema(
         action_log=action_log,
         mutations_observed=mutations_observed,
         local_edits=local_edits,
+        validation_session=validation_session,
     )
     extra: list[dict[str, str]] = []
     for rule in _SCHEMA_RULES:

review_proofs.py

@@ -3622,18 +3622,33 @@ def assess_work_issue_mode_isolation(report_text: str) -> dict:
     }
 
 
+def assess_work_issue_duplicate_prevention_report(report_text, **kwargs):
+    """#400: work-issue reports must classify duplicate-work prevention."""
+    from author_duplicate_work_gate import (
+        assess_work_issue_duplicate_prevention_report as _assess,
+    )
+
+    return _assess(report_text, **kwargs)
+
+
 def assess_work_issue_final_report(report_text: str) -> dict:
     """#139: composite verifier for work-issue final reports."""
     checks = {
         "workflow_source": assess_work_issue_workflow_source(report_text),
         "mode_isolation": assess_work_issue_mode_isolation(report_text),
+        "duplicate_prevention": assess_work_issue_duplicate_prevention_report(
+            report_text
+        ),
     }
 
     reasons = []
     downgraded = False
     for name, result in checks.items():
         verdict = result.get("verdict")
-        if verdict in ("missing", "incomplete"):
+        if result.get("block"):
+            downgraded = True
+            reasons.extend(result.get("reasons") or [])
+        elif verdict in ("missing", "incomplete"):
             downgraded = True
             reasons.extend(result.get("reasons") or [])
         elif result.get("downgraded") or not result.get("complete", True):
@@ -3641,6 +3656,9 @@ def assess_work_issue_final_report(report_text: str) -> dict:
             reasons.extend(
                 f"{name}: {r}" for r in (result.get("reasons") or [])
             )
+        elif result.get("proven") is False:
+            downgraded = True
+            reasons.extend(result.get("reasons") or [])
 
     grade = "A" if not downgraded else "downgraded"
     return {
@@ -5503,6 +5521,15 @@ def assess_validation_integrity_report(report_text, **kwargs):
     return _assess(report_text, **kwargs)
 
 
+def assess_validation_failure_history_report(report_text, **kwargs):
+    """#396: final reports must account for transient validation failures."""
+    from reviewer_validation_failure_history import (
+        assess_validation_failure_history_report as _assess,
+    )
+
+    return _assess(report_text, **kwargs)
+
+
 def assess_already_landed_classification_report(report_text, **kwargs):
     """#295: already-landed PRs are reconciliation-only, not review eligible."""
     from reviewer_already_landed_classification import (

reviewer_validation_failure_history.py

@@ -0,0 +1,198 @@
+"""Transient validation failure history verifier (#396).
+
+Reviewer sessions may observe validation failures that later pass on rerun.
+Final reports must document every failure observed during the session, not
+only the last passing result.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+_SECTION_RE = re.compile(
+    r"validation failure history",
+    re.IGNORECASE,
+)
+_FAILURE_ENTRY_RE = re.compile(
+    r"(?:failure\s*(?:#|entry)?\s*\d+|validation failure)\s*:",
+    re.IGNORECASE,
+)
+_COMMAND_RE = re.compile(
+    r"(?:command|validation command)\s*:\s*(.+)",
+    re.IGNORECASE,
+)
+_FAILING_TEST_RE = re.compile(
+    r"(?:failing test|failure|error)\s*:\s*(.+)",
+    re.IGNORECASE,
+)
+_CAUSE_RE = re.compile(
+    r"(?:suspected cause|cause)\s*:\s*(.+)",
+    re.IGNORECASE,
+)
+_REPRODUCED_RE = re.compile(
+    r"(?:reproduced|reproduces)\s*:\s*(yes|no|unknown)",
+    re.IGNORECASE,
+)
+_BASELINE_RE = re.compile(
+    r"(?:on baseline master|baseline master|exists on baseline)\s*:\s*(yes|no|unknown|not checked)",
+    re.IGNORECASE,
+)
+_PR_CAUSED_RE = re.compile(
+    r"(?:pr[- ]caused|pr caused)\s*:\s*(yes|no|unknown|not proven)",
+    re.IGNORECASE,
+)
+_TRANSIENT_STATUS_RE = re.compile(
+    r"(?:passed after transient failure investigation|transient failure investigation)",
+    re.IGNORECASE,
+)
+_PLAIN_PASS_RE = re.compile(
+    r"validation\s*:\s*(?:pass|passed|strong|ok|green)\b",
+    re.IGNORECASE,
+)
+_ENV_CLEANUP_RE = re.compile(
+    r"(?:environmental cleanup|state cleaned|what changed between runs)\s*:",
+    re.IGNORECASE,
+)
+_UNKNOWN_CAUSE_RE = re.compile(
+    r"(?:suspected cause|cause)\s*:\s*(?:unknown|unexplained|not determined)",
+    re.IGNORECASE,
+)
+
+
+def _failure_documented_in_text(text: str, failure: dict[str, Any]) -> bool:
+    """Return True when *failure* appears documented in free-form report text."""
+    command = (failure.get("command") or "").strip()
+    failing = (failure.get("failing_test") or failure.get("error") or "").strip()
+    if command and command not in text:
+        return False
+    if failing and failing not in text:
+        return False
+    return bool(command or failing)
+
+
+def _section_has_structured_fields(text: str) -> bool:
+    if not _SECTION_RE.search(text):
+        return False
+    section_start = _SECTION_RE.search(text).start()
+    section = text[section_start:]
+    has_command = bool(_COMMAND_RE.search(section))
+    has_failure = bool(_FAILING_TEST_RE.search(section))
+    return has_command and has_failure
+
+
+def assess_validation_failure_history_report(
+    report_text: str,
+    *,
+    validation_session: dict | None = None,
+) -> dict[str, Any]:
+    """Require final reports to account for transient validation failures (#396)."""
+    text = report_text or ""
+    session = dict(validation_session or {})
+    reasons: list[str] = []
+    violations: list[str] = []
+
+    observed = list(session.get("observed_failures") or [])
+    final_status = (session.get("final_validation_status") or "").strip().lower()
+    cause_unknown = any(
+        (f.get("suspected_cause") or "").strip().lower() in {
+            "unknown", "unexplained", "not determined", ""
+        }
+        and not (f.get("pr_caused") or "").strip().lower() in {"yes", "no"}
+        for f in observed
+    ) or bool(session.get("cause_unknown"))
+
+    if not observed:
+        return {
+            "proven": True,
+            "block": False,
+            "reasons": [],
+            "violations": [],
+            "observed_failure_count": 0,
+            "safe_next_action": "proceed",
+        }
+
+    documented = _section_has_structured_fields(text)
+    if not documented:
+        for failure in observed:
+            if _failure_documented_in_text(text, failure):
+                documented = True
+                break
+
+    if not documented:
+        violations.append(
+            "session observed validation failure(s) but final report omits "
+            "Validation failure history"
+        )
+        reasons.append(
+            "every validation failure observed during the session must appear "
+            "in a Validation failure history section"
+        )
+
+    if documented and not _SECTION_RE.search(text):
+        reasons.append(
+            "failure details must appear under an explicit "
+            "'Validation failure history' heading"
+        )
+
+    for idx, failure in enumerate(observed, start=1):
+        prefix = f"failure #{idx}"
+        if not _failure_documented_in_text(text, failure) and documented:
+            reasons.append(
+                f"{prefix}: command and failing test/error not documented in report"
+            )
+        command = (failure.get("command") or "").strip()
+        failing = (failure.get("failing_test") or failure.get("error") or "").strip()
+        if not command:
+            reasons.append(f"{prefix}: missing command in session failure record")
+        if not failing:
+            reasons.append(
+                f"{prefix}: missing failing test or error in session failure record"
+            )
+
+    plain_pass = bool(_PLAIN_PASS_RE.search(text))
+    transient_wording = bool(_TRANSIENT_STATUS_RE.search(text))
+    final_passed = final_status in {
+        "passed",
+        "pass",
+        "passed_after_transient_failure_investigation",
+    }
+
+    if (final_passed or plain_pass) and observed:
+        if cause_unknown and not transient_wording:
+            violations.append(
+                "unknown transient failure cause cannot be erased as plain 'passed'"
+            )
+            reasons.append(
+                "when cause is unknown, use status "
+                "'passed after transient failure investigation'"
+            )
+        elif not transient_wording and final_status != "passed_after_transient_failure_investigation":
+            if not _ENV_CLEANUP_RE.search(text):
+                reasons.append(
+                    "later passing rerun must document what changed between runs "
+                    "or environmental cleanup performed"
+                )
+
+    if session.get("environmental_contamination") and not _ENV_CLEANUP_RE.search(text):
+        reasons.append(
+            "environmental /tmp state contamination must document cleanup or "
+            "what changed before the passing rerun"
+        )
+
+    proven = not reasons and not violations
+    return {
+        "proven": proven,
+        "block": bool(violations) or not proven,
+        "reasons": reasons,
+        "violations": violations,
+        "observed_failure_count": len(observed),
+        "documented": documented,
+        "safe_next_action": (
+            "add Validation failure history with command, failing test, cause, "
+            "reproduction, baseline comparison, and PR-caused evidence; use "
+            "'passed after transient failure investigation' when appropriate"
+            if not proven
+            else "proceed"
+        ),
+    }

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

@@ -538,6 +538,36 @@ Official validation integrity status must be one of:
 
 Do not invent a softer status.
 
+## 21A. Validation failure history rule
+
+Every validation failure observed during the review session must appear in the
+final report, even if a later rerun passes.
+
+Before claiming `Validation: passed`, list every failed validation command from
+the session under `Validation failure history`.
+
+For each failure, report:
+
+* command
+* failing test or error
+* suspected cause
+* whether it reproduced
+* whether it exists on baseline master
+* evidence for whether it is or is not PR-caused
+
+If a later rerun passes, also report:
+
+* what changed between runs
+* whether environmental state was cleaned (for example `/tmp` lock files)
+* why the pass is trustworthy
+
+If the cause is unknown, do not erase the earlier failure with plain
+`Validation: passed`. Use a status such as
+`passed after transient failure investigation`.
+
+`gitea_validate_review_final_report` rejects reports that omit known earlier
+validation failures when `validation_session.observed_failures` is supplied.
+
 ## 22. Baseline validation rule
 
 Do not run tests in the main checkout.
@@ -1096,6 +1126,7 @@ Controller Handoff:
 * Baseline worktree path:
 * Files reviewed:
 * Validation:
+* Validation failure history:
 * Official validation integrity status:
 * Terminal review mutation:
 * Review decision:

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

@@ -277,6 +277,24 @@ Do not select an issue based only on memory from a previous session.
 
 Before claiming or working on an issue, check whether there is already an open PR, branch, or active claim for that issue.
 
+Run `gitea_assess_author_duplicate_work` at these stages and stop when `block` is true:
+
+* `claim` — before `gitea_mark_issue`
+* `lock` — before `gitea_lock_issue`
+* `worktree` / `edit` — before creating a worktree or editing files
+* `commit` — immediately before `git commit`
+* `push` — immediately before `git push`
+* `create_pr` — immediately before `gitea_create_pr` (also enforced server-side)
+
+`gitea_mark_issue`, `gitea_lock_issue`, and `gitea_create_pr` enforce the same gate server-side and fail closed.
+
+If a concurrent open PR appears after work begins:
+
+* before commit or push — stop and preserve local work without pushing
+* after push but before PR creation — produce a reconciliation handoff instead of opening a PR
+
+Final reports must name the duplicate-work outcome (`duplicate PR prevented`, `duplicate branch prevented`, `duplicate commit prevented`, `duplicate push prevented`, `duplicate work not prevented`, or `reconciliation handoff`).
+
 If an open PR already exists for the issue, do not implement duplicate work.
 
 Classify the issue as:

tests/test_author_duplicate_work_gate.py

@@ -0,0 +1,102 @@
+"""Tests for early author duplicate-work gate (#400)."""
+import sys
+import unittest
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
+
+from author_duplicate_work_gate import (  # noqa: E402
+    ELIGIBILITY_OPEN_PR_EXISTS,
+    assess_author_duplicate_work,
+    assess_work_issue_duplicate_prevention_report,
+)
+
+
+def _open_pr(number: int = 397, issue: int = 395) -> dict:
+    return {
+        "number": number,
+        "head": {"ref": f"feat/issue-{issue}-example"},
+        "title": f"feat: example (Closes #{issue})",
+        "body": f"Closes #{issue}",
+    }
+
+
+class TestAuthorDuplicateWorkGate(unittest.TestCase):
+    def test_clear_when_no_duplicates(self):
+        result = assess_author_duplicate_work(
+            400,
+            stage="claim",
+            open_prs=[],
+            branch_names=["master", "feat/issue-399-other"],
+        )
+        self.assertTrue(result["allowed"])
+        self.assertFalse(result["block"])
+
+    def test_open_pr_blocks_claim(self):
+        result = assess_author_duplicate_work(
+            395,
+            stage="claim",
+            open_prs=[_open_pr()],
+            branch_names=[],
+        )
+        self.assertFalse(result["allowed"])
+        self.assertEqual(result["eligibility_class"], ELIGIBILITY_OPEN_PR_EXISTS)
+
+    def test_matching_branch_blocks_lock_not_create_pr(self):
+        branches = ["feat/issue-400-early-duplicate-work-gate"]
+        lock = assess_author_duplicate_work(
+            400,
+            stage="lock",
+            open_prs=[],
+            branch_names=branches,
+        )
+        self.assertFalse(lock["allowed"])
+
+        create_pr = assess_author_duplicate_work(
+            400,
+            stage="create_pr",
+            open_prs=[],
+            branch_names=branches,
+            matching_branches=[],
+        )
+        self.assertTrue(create_pr["allowed"])
+
+    def test_open_pr_blocks_create_pr_stage(self):
+        result = assess_author_duplicate_work(
+            395,
+            stage="create_pr",
+            open_prs=[_open_pr()],
+            branch_names=["feat/issue-395-proof-backed-review-handoff"],
+        )
+        self.assertFalse(result["allowed"])
+        self.assertEqual(result["outcome"], "duplicate_pr_prevented")
+
+    def test_push_stage_blocks_on_concurrent_pr(self):
+        result = assess_author_duplicate_work(
+            395,
+            stage="push",
+            open_prs=[_open_pr()],
+            branch_names=[],
+        )
+        self.assertFalse(result["allowed"])
+        self.assertEqual(result["outcome"], "duplicate_push_prevented")
+
+    def test_duplicate_prevention_report_requires_outcome(self):
+        bad = assess_work_issue_duplicate_prevention_report(
+            "Duplicate work detected for issue #395."
+        )
+        self.assertFalse(bad["proven"])
+
+        good = assess_work_issue_duplicate_prevention_report(
+            "Duplicate PR prevented; reconciliation handoff produced."
+        )
+        self.assertTrue(good["proven"])
+
+    def test_exported_from_review_proofs(self):
+        from review_proofs import assess_work_issue_duplicate_prevention_report as exported
+
+        self.assertTrue(callable(exported))
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_delete_branch_capability.py

@@ -0,0 +1,217 @@
+"""Tests for gitea_delete_branch capability gate (Issue #408).
+
+``gitea_delete_branch`` requires the exact ``gitea.branch.delete`` operation:
+without it the delete fails closed (no preflight, no auth lookup, no API call,
+structured permission report). With it, deletion proceeds through existing
+preflight and audit unchanged.
+"""
+import json
+import os
+import sys
+import tempfile
+import unittest
+from unittest.mock import patch
+
+sys.path.insert(0, str(__import__("pathlib").Path(__file__).resolve().parent.parent))
+
+import mcp_server
+from mcp_server import gitea_delete_branch
+
+FAKE_AUTH = "token fake"
+
+AUTHOR_NO_DELETE = {
+    "profile_name": "prgs-author",
+    "allowed_operations": [
+        "gitea.read", "gitea.pr.create", "gitea.pr.comment",
+        "gitea.branch.push", "gitea.issue.comment",
+    ],
+    "forbidden_operations": ["gitea.pr.approve", "gitea.pr.merge"],
+    "audit_label": "prgs-author",
+}
+
+AUTHOR_WITH_DELETE = {
+    "profile_name": "prgs-author-deleter",
+    "allowed_operations": AUTHOR_NO_DELETE["allowed_operations"] + [
+        "gitea.branch.delete",
+    ],
+    "forbidden_operations": ["gitea.pr.approve", "gitea.pr.merge"],
+    "audit_label": "prgs-author-deleter",
+}
+
+CONFIG = {
+    "version": 2,
+    "contexts": {
+        "ctx": {
+            "enabled": True,
+            "gitea": {"enabled": True, "base_url": "https://gitea.example.com"},
+        }
+    },
+    "profiles": {
+        "author-no-delete": {
+            "enabled": True,
+            "context": "ctx",
+            "role": "author",
+            "username": "author-user",
+            "auth": {"type": "env", "name": "GITEA_TOKEN_AUTHOR"},
+            "allowed_operations": AUTHOR_NO_DELETE["allowed_operations"],
+            "forbidden_operations": [],
+            "execution_profile": "author-no-delete",
+        },
+        "author-with-delete": {
+            "enabled": True,
+            "context": "ctx",
+            "role": "author",
+            "username": "deleter-user",
+            "auth": {"type": "env", "name": "GITEA_TOKEN_AUTHOR"},
+            "allowed_operations": AUTHOR_WITH_DELETE["allowed_operations"],
+            "forbidden_operations": [],
+            "execution_profile": "author-with-delete",
+        },
+        "reviewer-profile": {
+            "enabled": True,
+            "context": "ctx",
+            "role": "reviewer",
+            "username": "reviewer-user",
+            "auth": {"type": "env", "name": "GITEA_TOKEN_REVIEWER"},
+            "allowed_operations": [
+                "gitea.read", "gitea.pr.review", "gitea.pr.merge",
+            ],
+            "forbidden_operations": [
+                "gitea.branch.delete", "gitea.branch.push", "gitea.pr.create",
+            ],
+            "execution_profile": "reviewer-profile",
+        },
+    },
+    "rules": {"allow_runtime_switching": False},
+}
+
+
+class TestDeleteBranchToolGate(unittest.TestCase):
+
+    def setUp(self):
+        self._preflight_snapshot = (
+            mcp_server._preflight_whoami_called,
+            mcp_server._preflight_capability_called,
+        )
+        mcp_server._preflight_whoami_called = False
+        mcp_server._preflight_capability_called = False
+        self._remotes = patch.dict(mcp_server.REMOTES, {
+            "prgs": {"host": "gitea.example.com", "org": "Example-Org",
+                     "repo": "Example-Repo"},
+        })
+        self._remotes.start()
+        mcp_server._IDENTITY_CACHE.clear()
+        patch("gitea_config.load_config", return_value={}).start()
+        patch(
+            "gitea_config.is_runtime_switching_enabled", return_value=False
+        ).start()
+        patch("gitea_audit.audit_enabled", return_value=False).start()
+        self.mock_api = patch("mcp_server.api_request").start()
+        self.mock_auth = patch(
+            "mcp_server.get_auth_header", return_value=FAKE_AUTH
+        ).start()
+
+    def tearDown(self):
+        patch.stopall()
+        mcp_server._IDENTITY_CACHE.clear()
+        mcp_server._preflight_whoami_called, mcp_server._preflight_capability_called = (
+            self._preflight_snapshot
+        )
+
+    def _set_profile(self, profile):
+        patch("mcp_server.get_profile", return_value=profile).start()
+
+    def test_blocked_without_delete_capability(self):
+        self._set_profile(AUTHOR_NO_DELETE)
+        res = gitea_delete_branch(branch="feat/branch", remote="prgs")
+        self.assertFalse(res["success"])
+        self.assertFalse(res["performed"])
+        self.assertEqual(res["required_permission"], "gitea.branch.delete")
+        self.assertTrue(res["reasons"])
+        self.assertEqual(
+            res["permission_report"]["missing_permission"],
+            "gitea.branch.delete",
+        )
+        self.mock_api.assert_not_called()
+        self.mock_auth.assert_not_called()
+
+    def test_allowed_delete_proceeds(self):
+        self._set_profile(AUTHOR_WITH_DELETE)
+        mcp_server.record_preflight_check("whoami")
+        mcp_server.record_preflight_check("capability", resolved_role="author")
+        res = gitea_delete_branch(branch="feat/branch", remote="prgs")
+        self.assertTrue(res["success"])
+        self.assertIn("deleted", res["message"])
+        delete_calls = [
+            c for c in self.mock_api.call_args_list if c.args[0] == "DELETE"
+        ]
+        self.assertTrue(delete_calls)
+
+    def test_allowed_delete_audited_with_capability_proof(self):
+        self._set_profile(AUTHOR_WITH_DELETE)
+        patch("gitea_audit.audit_enabled", return_value=True).start()
+        mock_write = patch("gitea_audit.write_event").start()
+        mcp_server.record_preflight_check("whoami")
+        mcp_server.record_preflight_check("capability", resolved_role="author")
+        self.mock_api.return_value = {}
+        gitea_delete_branch(branch="feat/branch", remote="prgs")
+        mock_write.assert_called()
+        event = mock_write.call_args[0][0]
+        self.assertEqual(event["action"], "delete_branch")
+        self.assertEqual(
+            event["request_metadata"]["required_permission"],
+            "gitea.branch.delete",
+        )
+
+
+class TestDeleteBranchResolverParity(unittest.TestCase):
+    def setUp(self):
+        self._remotes = patch.dict(mcp_server.REMOTES, {
+            "prgs": {"host": "gitea.example.com", "org": "Example-Org",
+                     "repo": "Example-Repo"},
+        })
+        self._remotes.start()
+        mcp_server._IDENTITY_CACHE.clear()
+        self._dir = tempfile.TemporaryDirectory()
+        self.config_path = os.path.join(self._dir.name, "profiles.json")
+        with open(self.config_path, "w", encoding="utf-8") as fh:
+            fh.write(json.dumps(CONFIG))
+        patch(
+            "gitea_config.is_runtime_switching_enabled", return_value=False
+        ).start()
+        patch("gitea_audit.audit_enabled", return_value=False).start()
+        self.mock_api = patch("mcp_server.api_request").start()
+        patch("mcp_server.get_auth_header", return_value=FAKE_AUTH).start()
+
+    def tearDown(self):
+        patch.stopall()
+        mcp_server._IDENTITY_CACHE.clear()
+        self._dir.cleanup()
+
+    def _env(self, profile: str) -> dict:
+        return {
+            "GITEA_MCP_CONFIG": self.config_path,
+            "GITEA_MCP_PROFILE": profile,
+            "GITEA_TOKEN_AUTHOR": "author-pass",
+            "GITEA_TOKEN_REVIEWER": "reviewer-pass",
+        }
+
+    def test_reviewer_resolver_denial_blocks_raw_tool(self):
+        with patch.dict(os.environ, self._env("reviewer-profile"), clear=True):
+            resolve = mcp_server.gitea_resolve_task_capability(
+                task="delete_branch", remote="prgs")
+            self.assertFalse(resolve["allowed_in_current_session"])
+            res = gitea_delete_branch(branch="feat/branch", remote="prgs")
+        self.assertFalse(res["success"])
+        self.assertEqual(
+            res["permission_report"]["missing_permission"],
+            resolve["required_operation_permission"],
+        )
+        delete_calls = [
+            c for c in self.mock_api.call_args_list if c.args[0] == "DELETE"
+        ]
+        self.assertFalse(delete_calls)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_llm_workflow_split.py

@@ -95,6 +95,12 @@ def test_create_issue_workflow_contract():
     assert "## 9. Duplicate search before mutation" in text
 
 
+def test_author_duplicate_work_gate_exported():
+    from review_proofs import assess_work_issue_duplicate_prevention_report
+
+    assert callable(assess_work_issue_duplicate_prevention_report)
+
+
 def test_work_issue_workflow_contract():
     text = (SKILL_DIR / "workflows" / "work-issue.md").read_text(encoding="utf-8")
     assert "canonical: true" in text
@@ -207,6 +213,12 @@ def test_validation_integrity_verifier_exported():
     assert callable(assess_validation_integrity_report)
 
 
+def test_validation_failure_history_verifier_exported():
+    from review_proofs import assess_validation_failure_history_report
+
+    assert callable(assess_validation_failure_history_report)
+
+
 def test_prior_blocker_skip_verifier_exported():
     from review_proofs import assess_prior_blocker_skip_proof
 

tests/test_mcp_server.py

@@ -1005,9 +1005,19 @@ class TestReviewPR(unittest.TestCase):
 # ---------------------------------------------------------------------------
 class TestDeleteBranch(unittest.TestCase):
 
+    DELETE_PROFILE = {
+        "profile_name": "test-deleter",
+        "allowed_operations": ["gitea.read", "gitea.branch.delete"],
+        "forbidden_operations": [],
+        "audit_label": "test-deleter",
+    }
+
+    @patch("mcp_server.get_profile", return_value=DELETE_PROFILE)
     @patch("mcp_server.api_request")
     @patch("mcp_server.get_auth_header", return_value=FAKE_AUTH)
-    def test_delete_branch(self, _auth, mock_api):
+    def test_delete_branch(self, _auth, mock_api, _profile):
+        mcp_server.record_preflight_check("whoami")
+        mcp_server.record_preflight_check("capability", resolved_role="author")
         mock_api.return_value = {}
         result = gitea_delete_branch(branch="feat/branch")
         self.assertTrue(result["success"])

tests/test_reviewer_validation_failure_history.py

@@ -0,0 +1,147 @@
+"""Tests for transient validation failure history verifier (#396)."""
+import sys
+import unittest
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
+
+from final_report_validator import assess_final_report_validator  # noqa: E402
+from reviewer_validation_failure_history import (  # noqa: E402
+    assess_validation_failure_history_report,
+)
+
+
+def _full_history_report() -> str:
+    return "\n".join([
+        "Validation failure history",
+        "Failure #1:",
+        "Command: venv/bin/python -m pytest tests/test_worktrees.py -q",
+        "Failing test: tests/test_worktrees.py::TestWorktreeStart::test_rejects_untraceable_branches",
+        "Suspected cause: stale /tmp/gitea_issue_lock.json from prior session",
+        "Reproduced: no",
+        "On baseline master: no",
+        "PR-caused: no",
+        "What changed between runs: removed /tmp/gitea_issue_lock.json",
+        "Environmental cleanup: lock file removed before rerun",
+        "Validation: passed after transient failure investigation",
+        "Final validation command: venv/bin/python -m pytest tests/ -q",
+        "Final result: 1497 passed, 6 skipped",
+    ])
+
+
+class TestValidationFailureHistory(unittest.TestCase):
+    def test_no_failures_observed_passes(self):
+        result = assess_validation_failure_history_report(
+            "Validation: passed",
+            validation_session={"observed_failures": []},
+        )
+        self.assertTrue(result["proven"])
+
+    def test_omitted_failure_blocks(self):
+        result = assess_validation_failure_history_report(
+            "Validation: passed\n1497 passed, 6 skipped",
+            validation_session={
+                "observed_failures": [{
+                    "command": "pytest tests/test_worktrees.py -q",
+                    "failing_test": (
+                        "tests/test_worktrees.py::TestWorktreeStart::"
+                        "test_rejects_untraceable_branches"
+                    ),
+                }],
+                "final_validation_status": "passed",
+            },
+        )
+        self.assertFalse(result["proven"])
+        self.assertTrue(result["block"])
+
+    def test_full_history_with_transient_investigation_passes(self):
+        result = assess_validation_failure_history_report(
+            _full_history_report(),
+            validation_session={
+                "observed_failures": [{
+                    "command": (
+                        "venv/bin/python -m pytest tests/test_worktrees.py -q"
+                    ),
+                    "failing_test": (
+                        "tests/test_worktrees.py::TestWorktreeStart::"
+                        "test_rejects_untraceable_branches"
+                    ),
+                    "suspected_cause": "stale /tmp/gitea_issue_lock.json",
+                    "reproduced": "no",
+                    "on_baseline_master": "no",
+                    "pr_caused": "no",
+                }],
+                "final_validation_status": "passed_after_transient_failure_investigation",
+                "environmental_contamination": True,
+            },
+        )
+        self.assertTrue(result["proven"], result["reasons"])
+
+    def test_baseline_equivalent_failure_documented(self):
+        report = "\n".join([
+            "Validation failure history",
+            "Command: pytest tests/test_example.py -q",
+            "Failing test: tests/test_example.py::test_flaky",
+            "Suspected cause: pre-existing master failure",
+            "On baseline master: yes",
+            "PR-caused: no",
+            "What changed between runs: none; failure matches baseline",
+            "Validation: passed after transient failure investigation",
+        ])
+        result = assess_validation_failure_history_report(
+            report,
+            validation_session={
+                "observed_failures": [{
+                    "command": "pytest tests/test_example.py -q",
+                    "failing_test": "tests/test_example.py::test_flaky",
+                    "on_baseline_master": "yes",
+                    "pr_caused": "no",
+                }],
+                "final_validation_status": "passed_after_transient_failure_investigation",
+            },
+        )
+        self.assertTrue(result["proven"], result["reasons"])
+
+    def test_unknown_cause_plain_pass_blocks(self):
+        result = assess_validation_failure_history_report(
+            "Validation: passed",
+            validation_session={
+                "observed_failures": [{
+                    "command": "pytest tests/ -q",
+                    "failing_test": "tests/test_foo.py::test_bar",
+                    "suspected_cause": "unknown",
+                }],
+                "final_validation_status": "passed",
+                "cause_unknown": True,
+            },
+        )
+        self.assertFalse(result["proven"])
+        self.assertTrue(any("transient" in r.lower() for r in result["reasons"]))
+
+    def test_final_report_validator_rejects_omitted_failure(self):
+        result = assess_final_report_validator(
+            "Validation: passed",
+            "review_pr",
+            validation_session={
+                "observed_failures": [{
+                    "command": "pytest tests/ -q",
+                    "failing_test": "tests/test_foo.py::test_bar",
+                }],
+            },
+        )
+        self.assertTrue(result["blocked"] or result["downgraded"])
+        self.assertTrue(
+            any(
+                f.get("rule_id") == "reviewer.validation_failure_history"
+                for f in result.get("findings") or []
+            )
+        )
+
+    def test_exported_from_review_proofs(self):
+        from review_proofs import assess_validation_failure_history_report as exported
+
+        self.assertTrue(callable(exported))
+
+
+if __name__ == "__main__":
+    unittest.main()