Clarify that Policy is live (not a Phase 1 placeholder) in shell docs and
nav module docstring. Strengthen the Policy nav graduation test to assert
status=live and absence of the nav-stub CSS class.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Address review blockers (review #555):
- Graduate /policy nav item out of STUB_PAGES to live nav in webui/nav.py
- Update docs/webui-local-dev.md to list /policy as a live surface and remove stub copy
- Run console_redaction on HTML emit path in webui/policy_views.py before rendering HTML
- Add planted-secret HTML redaction test and nav graduation test in tests/test_webui_policy_visibility.py
Resolve the #638 shell landing against the #639 dashboard:
- webui/layout.py: drop the flat NAV_ITEMS tuple in favor of master's
grouped NAV_GROUPS nav-config module.
- webui/nav.py: register /system-health as a live item in the Health
group, satisfying issue #639 AC5 through the canonical nav source.
- docs/webui-local-dev.md: keep both additive sections (#638 shell and
#639 dashboard).
- tests/test_webui_system_health_dashboard.py: assert the nav entry via
iter_nav_items() instead of the removed NAV_ITEMS tuple.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Review #526 found that event_type reached the serialized timeline payload
without crossing the redaction/validation boundary every other free-text
field on the same event crosses. A synthetic secret-shaped 40-hex canary
was redacted through message but survived verbatim through event_type on
the same control-plane record.
CTH heading path: CTH_TYPES is now the single authority for what a CTH
type may be. canonical_thread_handoff.is_known_cth_type() is that
authority, used by format_cth_body (write), assess_cth_comment (assess),
and now the read path too; parse_cth_comment reports membership as
cth_type_known and stays total. adapt_cth_comments serializes
handoff:<type> only for a declared type and otherwise emits the constant
handoff:unrecognized, so arbitrary, malformed, secret-shaped, or
whitespace-manipulated heading content never becomes an event_type.
Control-plane path: a stored event_type is treated as source data.
_safe_cp_event_type accepts only an ordinary identifier that is not a
bare secret-shaped hex run and that a redaction pass leaves unchanged;
anything else fails closed to the constant unsafe:redacted and marks the
event sensitive. The value is never emitted verbatim, never partially
sanitized, and never rewritten into a different valid-looking type.
Remaining serialized-field audit: event_key ids must be plain numeric
identifiers, the CTH adapter refuses a scope it cannot express, per-source
failure reasons are redacted (they can quote an authenticated fetch error),
and echoed scope/filter values are guarded so reflection is not a bypass.
Legitimate values are preserved: every declared CTH type, the real
producer types (assigned, lease_released, lease_adopted,
dependency_edge_state_change, allocation, pr.opened, lease.renew), and
the existing issue, PR, evidence, and full-SHA references.
Tests seed the canary independently through both event_type paths with a
benign message, so message redaction cannot be why they pass; each
inspects event_type directly, asserts the canary is absent from the
complete serialized payload, and asserts scan_for_secrets finds nothing.
tests/test_webui_timeline.py 64 passed, 15 subtests
pytest -k webui 362 passed, 285 subtests
pytest -k redact 79 passed
tests/test_canonical_thread_handoff.py tests/test_control_plane_db.py 29 passed
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Remediates the two blocking findings in review #522 on PR #849.
F1 — the session filter dimension was dead end to end. No source could
produce an event carrying a session identifier, so filter_events dropped
every event whenever session was supplied and the API answered with a
green, empty page. An empty result reads to an operator as "no such
session activity", which is a stronger and false claim.
Each source now declares which filter dimensions its records can actually
carry. The control-plane events table is (event_id, work_item_id,
event_type, message, created_at) and records no session, so that source
declares the session dimension unsupported rather than pretending to
answer it; the dead read of a non-existent session_id column is removed.
A CTH handoff comment declares its own Session field, so the handoff
adapter populates session_id from that declared field — authoritative
source data, never inferred from an actor, work item, or message text.
When no source that ran can carry a requested dimension, load_timeline
refuses with ok=false and a structured error naming the unsupported
filters and the per-source reason, and the route answers 422. A source
that can answer the dimension and simply matched nothing still returns
200 with an honest empty page. Ordering, pagination, and the issue/PR
filters are unchanged.
F2 — evidence_refs bypassed redaction and could emit a credential
verbatim. proof and decision were passed to _extract_evidence_refs before
redaction, and the SHA pattern matched any 7-40 character lowercase hex
run, which is exactly the shape of a Gitea access token.
Redaction now runs first and every derived value is taken from the
redacted text. A commit reference is recognised only where the source
text declares one (commit, head, base, sha, ...), so an undeclared hex
run is never lifted out of prose into a structured field; this also drops
the ordinary-word noise the reviewer noted. Every reference is then
independently revalidated against an allowed shape and a second redaction
pass immediately before serialization, failing closed by dropping
anything unproven and flagging the event sensitive. Actor is redacted for
the same reason, and a secret-shaped session value is dropped rather than
emitted. Legitimate issue, PR, short-SHA and full 40-character SHA
references stay usable.
Tests: 16 added. Session filtering is now driven through the CTH adapter
and the composed load_timeline/API path rather than a hand-built
WorkflowEvent, covering a match, an honest empty result, pagination and
ordering under the filter, the 422 refusal, and the per-source support
declaration. Redaction coverage asserts a synthetic 40-character hex
value (not a real credential) appears nowhere in the complete serialized
payload including evidence_refs, that the independent revalidation drops
unproven references, and that legitimate references still resolve.
pytest tests/test_webui_timeline.py: 42 passed (was 26).
pytest -k webui: 340 passed, 270 subtests passed (was 324).
Full suite: 4607 passed, 12 failed, 6 skipped, 684 subtests passed — the
same 12 failures as the master baseline, none under webui/.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Phase 1 child of the Web Console epic #631. Adds the operator-facing
system-health dashboard on top of the read-only system-health API landed
by #634, so runtime problems are visible on a surface instead of being
discovered late through failed LLM sessions.
- webui/system_health_views.py (new): renders the SystemHealthSnapshot as
readiness, stale-runtime parity, version/uptime, dependency, MCP
namespace, probe-error, and recovery cards.
- webui/app.py: GET /system-health, sharing load_system_health() with the
JSON API so page and API cannot disagree. ?deep=1 behaves as on the API.
- webui/layout.py: nav entry and health card/badge styles.
- tests/test_webui_system_health_dashboard.py (new, 26 cases).
- docs/webui-local-dev.md: route, field authority, and redaction split.
Readiness honesty is preserved from the API: ready and readiness_complete
render separately, a probe that did not run is listed under "Not probed"
rather than counted healthy, and mutation safety is never claimed when the
runtime is stale or parity is indeterminate.
Redaction is split by field kind. Free text (probe details, reasons, probe
errors) passes through system_health.redact. Structured fields (commit
SHAs, probe names, statuses, timestamps) are HTML-escaped only: redact's
opaque-token rule matches any run of 32 or more characters, so routing a
40-character git SHA through it rendered "[redacted]" and blanked the
parity evidence the page exists to show.
Non-goals honored: no restart or reload controls (Phase 2, #642), no
manual process-kill guidance (#630). Read-only throughout.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Phase 1 child of the Web Console epic #631. Adds a durable, versioned
WorkflowEvent schema with per-source adapters and a read-only query API so
operators can browse a unified timeline of workflow events, decisions, tool
calls, and handoffs instead of scattered evidence.
- webui/timeline.py (new): versioned WorkflowEvent schema; control-plane
event adapter and Gitea CTH handoff-comment adapter; read-only mode=ro
control-plane reader; conjunctive filter by issue/PR/session; stable
(timestamp, source_rank, event_key) ordering; bounded pagination;
fail-soft per-source status; redaction at the boundary, fail closed.
- webui/app.py: GET /api/v1/timeline read-only route with thread-scoped,
fail-soft handoff comment source.
- tests/test_webui_timeline.py (new): schema, adapters, redaction of
secret-like payloads, filter/sort/pagination, scoped CP reader,
fail-soft composition, and API integration.
- docs/webui-local-dev.md: timeline route and field-authority notes.
Read-only Phase 1; no mutation of historical events; no full chat replay;
no unredacted tool-argument storage.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Phase 3 child of the Web Console epic #631. Operators can now see the active
workflow policy/guardrail configuration from the console instead of reading the
repo tree.
- webui/policy_inventory.py (new): redacted, machine-readable guardrail
inventory. One row per major guardrail (role separation/RBAC, lease rules,
worktree binding, merge confirmation, redaction, contamination, allocator
policy, audit logging, mutation gating) with source pointers (file/module/doc)
and a compact active projection from the existing safe policy accessors.
Fail-soft per entry; whole payload run through console_redaction before emit;
diff vs documented default where feasible.
- webui/policy_views.py (new): HTML cards with source pointers, active config,
and the documented-default diff; read-only page copy, no forms.
- webui/app.py: register GET /policy and GET /api/v1/policy (additive).
- webui/layout.py: add Policy nav item.
- tests/test_webui_policy_visibility.py (new): guardrail presence + source
pointers (AC1), redaction incl. planted-secret masking and scan_for_secrets
(AC2/AC3), read-only page + no-mutation routes (AC4), fail-soft rendering.
- docs/webui-local-dev.md: route table + read-only policy-visibility section.
Read-only throughout; no policy editing, no gate-weakening toggle, secrets
redacted.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Resolve conflict remediation for PR #818 (Closes#638) against master
caaae9b6. Two conflicts in webui/app.py, both resolved as unions since
the Phase 1 shell work (#638) and the merged master changes touch
disjoint concerns:
- Imports: keep the new webui.nav (NAV_GROUPS, STUB_PAGES) import from
#638 alongside master's expanded project_registry / project_views API
(ProjectRegistry, RegistryError, known_project_ids,
project_detail_to_dict, render_registry_error).
- Route table: keep master's read-only /api/console/security-model route
alongside #638's read-only Phase 1 stub routes (STUB_PAGES).
No behavior change beyond union; console stays read-only. docs/webui-local-dev.md
auto-merged. Full webui suite green (272 passed, 310 subtests).
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Introduce a single nav-config module (webui/nav.py) driving grouped
navigation across the epic #631 Phase 1 information architecture:
Health, Traffic, Runtime/Sessions, Projects, Inventory, Timeline,
Policy (placeholder), and Insights (placeholder). The shell header now
carries a read-only environment badge (local/remote from WEBUI_HOST), a
mode: read-only badge, and a Docs link. Home summarizes the console
purpose, lists the Phase 1 surfaces by group, and links the MVP legacy
pages.
Not-yet-implemented surfaces (/sessions, /inventory, /timeline,
/policy, /insights) resolve to graceful read-only stub pages rather than
404s; their backing views land in later child issues of #631 (inventory
surfaces backed by #636). Mutating methods on stub routes still fail
closed with read-only-mvp. No privileged action controls are added.
Adds tests/test_webui_shell.py covering nav groups, badge presence,
environment classification, docs link, stub routes (200 + read-only),
and home content. Updates docs/webui-local-dev.md with the new routes
and a Phase 1 shell section.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Adds `GET /api/v1/system/health`, a structured read-only health surface for
automated readiness checks, and keeps `/health` as the cheap liveness probe.
webui/system_health.py composes a DTO from fail-soft dependency probes: the
control-plane database, the local checkout, and — opt-in via `?deep=1` — live
Gitea reachability, each carrying status, reason, and probe latency. Required
probes drive readiness; the optional Gitea probe can only degrade overall
status, because local inventory stays serveable when the remote is
unreachable. A probe that did not run leaves readiness incomplete rather than
silently passing.
Read-only throughout: the control-plane database is opened through a `mode=ro`
URI because `ControlPlaneDB.__init__` creates directories and runs migrations,
which a health check must never do. No restart or reload control is exposed;
those are Phase 2 and #630 forbids process-kill recovery.
No unproven claims: `stale_runtime.mutation_safe` is true only when the
runtime, checkout, and remote commits are all known and equal, and MCP
namespaces always report `unproven` because a web process cannot exercise the
IDE-managed client path (#543). Probe details are redacted at the browser
boundary — URLs lose userinfo and query strings, credential-shaped text is
masked.
`/health` is expanded additively: every MVP key is retained, plus `started_at`,
`uptime_seconds`, and a pointer to the versioned API. The versioned route
returns 503 when not ready so automation can branch on the status code alone.
Verified at master 9eb0f29: focused file 40 passed / 11 subtests; `-k "webui or
health"` 230 passed / 159 subtests; full suite 4358 passed with the 11
pre-existing master-drift failures unchanged from the clean-master baseline.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Evolve the MVP project registry (#427) into a versioned, fail-closed
project registry API for the console (Phase 1, read-only).
- Add schema version 2 with project `status`, per-step onboarding
`state`/`required`, optional redacted `last_seen_health`, and
`remote_name`. Version 1 files stay loadable and are normalized with
explicit defaults.
- Serve `/api/v1/projects` and `/api/v1/projects/{project_id}` with API
provenance (`api_version`, `schema_version`, `source`). `/api/projects`
is retained as an unversioned Phase 1 alias.
- Replace bare `ValueError` with `RegistryError`, carrying an operator
`remediation` and `field_path`; invalid registries fail closed as a
500 JSON payload or a dedicated HTML error page instead of a traceback.
- Reject credential-shaped keys before any DTO is built, reusing
`registry_safety.is_forbidden_key` as the single source of truth
shared with the worker registry (#798).
- Render HTML views from `project_to_dict`, so the console and the JSON
API cannot disagree about status or onboarding progress.
- Document the contract in docs/webui-project-registry-api.md.
Tests: registry load/validate (valid, missing project, schema
validation, credential rejection) and API route coverage.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Phase 1 of the MCP Control Plane Web Console (#631) defines the model that
future gated writes must pass through, and enables none of them.
The read-only MVP (#426-#436) ships with no authentication; protection comes
from network placement alone (#435). That is adequate while every route is a
GET and inadequate the moment Phase 2 wires a write. This lands the authority
first, so no write can later be added without something to check it against.
webui/console_authz.py
Identity sources (none / local-dev / access-proxy), a four-role matrix
(viewer, operator, controller, admin), the privileged-action list, and a
fail-closed authorize(). Every action maps to a task_key in
task_capability_map, so the console cannot invent an authority the MCP layer
does not already define. Roles are always server-side configuration, never a
client assertion. Deny reasons are closed and enumerated; there is no
implicit allow branch, and even an allow reports execution_enabled=false
while ACTIVE_PHASE is 1.
webui/console_redaction.py
One redaction pass for API payloads, rendered HTML, logs, and audit records.
Reuses gitea_audit.redact as the shared authority rather than forking it,
then adds console patterns for keychain references, credential assignments,
PEM private-key blocks, and JWTs. Never raises: an unredactable value
degrades to the placeholder rather than being emitted raw.
webui/console_audit.py
Console-side audit records, which gitea_audit cannot supply: it records MCP
mutations and carries no console actor, identity source, correlation id, or
retention class, and an authorization denial is not a mutation at all. The
two are additive and join on correlation.request_id. Records are redacted at
build time, re-scanned at write time, and dropped rather than persisted if
they still trip a detector. Retention is per-record; an unknown action is
retained as privileged rather than standard.
webui/app.py
Attaches an authorization block to the existing preview and attempt routes
and records the decision. The terminal outcome is unchanged - gated_actions
still fails closed for every action - so this cannot loosen anything. Adds
GET /api/console/security-model publishing the three policies as JSON.
Probe authentication is deliberately declarative in this slice:
probe_auth_required() reports operator intent and no route consults it. The
documentation says so plainly and a regression test pins the not-enforced
status, so wiring it in Phase 2 is a deliberate change rather than a silent
one. An operator who sets the variable believing it protects a probe would be
worse off than one who knows it does not.
Tests: tests/test_webui_console_authz_audit.py - 75 passed, 93 subtests,
covering each acceptance criterion and each test the issue requires
(redaction units, default-deny for unauthenticated write stubs, audit record
creation for a simulated privileged preview).
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Add the declarative worker registry that epic #797 makes the source of
truth for the scheduled multi-LLM worker fleet.
Providers and configured workers are modelled as separate entities so a
provider can be listed with no worker configured, and so provider facts
are not copied into every worker record. A worker records provider,
model, project, role, namespace, profile, workflow, schedule, timeout,
enabled state, and scheduler metadata.
Validation fails closed: unknown fields are refused rather than ignored,
so a typo cannot silently disable a timeout; a worker naming an
undeclared provider is rejected; worker ids, provider ids, and
LaunchAgent labels must be unique.
Persistence is atomic (temp file in the same directory, fsync, replace).
Every superseded document is retained as a numbered revision, and
rollback republishes a chosen revision as a new head, so history stays
append-only and a rollback is itself reversible.
The credential-rejection guard is extracted to webui/registry_safety.py
so both registries share one implementation instead of two copies of a
security check; project_registry.py keeps identical behaviour.
Scope: data model, validation, persistence only. No routes, scheduler,
process control, or provider probing - those are #799/#800/#804/#805.
The workers array ships empty because populating it is #808.
Tests: tests/test_webui_worker_registry.py, 44 cases.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Add consolidated MVP route/read-only tests, path-filter helpers, and
scripts/test-webui plus scripts/ci-webui-check for Jenkins path-filtered
runs on PRs touching web UI surfaces.
Closes#436
Conflict resolution left webui/lease_loader.py importing ISSUE_LOCK_FILE
from merged_cleanup_reconcile (not exported); import from
issue_lock_provenance instead. Validate WEBUI_HOST before loading the
full app stack so 0.0.0.0 refuse exits immediately.
Addresses reviewer findings on PR #456 / issue #435.
Add bind-host assessment that refuses 0.0.0.0/:: without override,
warns on non-loopback binds, and documents internal-only MVP serving.
Health endpoint exposes deployment metadata; docs cover Access/VPN/WARP
and runtime env assumptions without embedding secrets in the client.
Closes#435
The conflict-resolution rebase incorrectly kept ISSUE_LOCK_FILE in the
merged_cleanup_reconcile import, but that symbol was moved to
issue_lock_provenance on master. This broke webui import collection.
Fixes the reviewer-reported regression on PR #448.
Adds read-only /runtime and /api/runtime routes showing active profile,
identity, config model, git sync vs remote master, shell health,
workflow/schema hashes, and stale-runtime warnings with #420 guidance.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
When Gitea credentials are missing or the queue fetch fails closed,
show "Not loaded" instead of "No open items." and keep pagination
marked unavailable. Successful empty inventories still show the empty
state with complete pagination proof.
Closes#458
Adds read-only /leases and /api/leases routes for issue claim inventory,
reviewer PR lease comments, duplicate PR/branch warnings, and local issue
lock visibility without acquire/release mutations.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Adds read-only /worktrees and /api/worktrees routes that scan branches/
directories, classify worktree state, flag missing preserved worktrees from
the issue lock, and surface the canonical cleanup prompt without deletion.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Expose /audit and /api/audit for local validator previews using
final_report_validator and review schema checks. Operators can paste
reports, auto-detect task kind, and get structured findings with
suggested next prompts and issue-comment drafts.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Expand queue dashboard tests for stale/duplicate badges, title-linked
issues, and nav coverage; drop unused import in queue_loader.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Adds read-only /queue and /api/queue routes that load open PRs and issues
from the default registry project via gitea_auth, surface pagination proof,
and classify items with claimed/blocked/in-review/duplicate badges.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Surface short copy/paste operator prompts derived from canonical workflow
files with workflow path and SHA-256 citations. Adds /prompts, /prompts/{id},
and /api/prompts with one-click copy.
Closes#428
Expose short copy/paste operator prompts on /prompts derived from canonical
workflow files with workflow path and SHA-256 hash. Adds JSON export at
/api/prompts, copy buttons, and tests.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
sysadminandClaude Opus 4.8
jcwalker3