Lapsed claims were reopened with no attempt ceiling, so a run whose worker
lost its lease (including a 409 before dispatch) could cycle forever.
Activity Core now reopens a lapsed claim below OPS_RUN_MAX_ATTEMPTS and sets
terminal 'expired' at the ceiling. The owner is kept for audit, and
/ops-runs/expire-leases reports reopened and expired counts. Answers the
REINAH-WP-0003-T05 open point on thread 27cd4c47 (ACTIVITY-WP-ADHOC-2026-10-19).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 244572@bnt-lap001
Assistant-Session: 3b235e1a-ae45-4d0f-ae70-17a9bd78ece8
ACTIVITY-WP-0046: grant v2 carries a publish scope (remote, branch, path
prefixes inside allowed_paths, fast-forward only) in its canonical grant id;
v1 stays local-only with unchanged ids.
ACTIVITY-WP-0047: disabled daily-security-flow definition dispatching the
the-custodian runner as an ops_run (CUST-WP-0079-T03).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 296847@bnt-lap001
Assistant-Session: 4f997d46-1b9d-4131-9699-0c353a215155
Install repo-manager e77f8b4 from its own uv.lock into an isolated venv in
the builder stage (commit identity asserted via git), copy it into the
runtime image and set ACTIVITY_CORE_RMGR_BIN. Add a pin test and the
founder-executed rollout package (Argo CD manifest diff, PVC, token route
blocker, trial runbooks). Nothing deployed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 692292@bnt-lap001
Assistant-Session: 89c7d8e8-e1b3-47eb-9411-a0ab227fca2f
Precondition 3 of the activation plan. New bounded operation
owner_return_pickup (state-hub-owner-return/pickup), executed by the existing
execute_bounded_operation activity and operator-triggered via the ops console
trigger with confirm_side_effect; no schedule, definition stays disabled.
The caller reads an operator-supplied C-25 checker artefact pinned by sha256
(the checker needs a local checkout and cannot run in the worker), requires a
full-SHA source_revision on the report and every C-25 finding, refuses failed
or incomplete collection, binds build_current_tasks to that SHA (Forge HEAD
verified), runs the State Hub planner vendored byte-identically from
state-hub 150cb0b, and submits only reviewable candidates under
ACTIVITY_CORE_OWNER_RETURN_PICKUP_ENABLED. Evidence is IDs/counts only.
build_current_tasks now refuses a report without source_revision and names
source_revision_reason when it is null.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 692292@bnt-lap001
Assistant-Session: 89c7d8e8-e1b3-47eb-9411-a0ab227fca2f
The repo-manager runner reports per-repository rows under results[], not
receipts[]; the projection counted nothing against the real output.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 692292@bnt-lap001
Assistant-Session: 89c7d8e8-e1b3-47eb-9411-a0ab227fca2f
Build the planner's current_tasks from immutable Forge commit content at the
C-25 scan SHA (no checkouts, ADR-002). Fails closed on revision mismatch,
non-head scans, missing/malformed fields and duplicate identities; unresolved
predecessors and decisions stay unknown (None).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 692292@bnt-lap001
Assistant-Session: 89c7d8e8-e1b3-47eb-9411-a0ab227fca2f
Disabled reviewed definition, pinned profile/grant, server-side unique
submission adapter with receipts, and opt-in activation plan. No schedule,
deployment or dispatch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 671493@bnt-lap001
Assistant-Session: 60d7b11f-78f9-4731-9367-918511b2762d
The openbao-activity-core store has used OpenBao Kubernetes auth since
2026-09-21 and synced cleanly for more than 24 hours. Remove the
openbao-eso-token-apply script and Makefile target, point the manifest
comments and runbooks at the Kubernetes-auth store, and guard against
reintroducing the script.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 151606@bnt-lap001
Assistant-Session: 3c0a4ad5-bb8b-4bf7-b9f0-fa5f29204e48
ACTIVITY_CORE_WORKERS maps worker_id=ENV_NAME, where each token env must be
ACTIVITY_CORE_WORKER_TOKEN[_SUFFIX]. Without the map, the legacy single pair
behaves exactly as before. Duplicate identities, missing or shared tokens, a
token equal to the operator token, and an unlisted legacy identity all fail
worker mutations closed with 503. Operator/SSO reads keep working.
Declare per-identity OpenBao paths and an ExternalSecret, not yet applied.
The policy, seeding and rollout are waiting tasks T02-T04, answering
secrets-engine SECRETS-WP-0009-T03 and SECRETS-WP-0011-T04.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 151606@bnt-lap001
Assistant-Session: 3c0a4ad5-bb8b-4bf7-b9f0-fa5f29204e48
The status surface queried only four fixed State Hub event types, so
bounded-operation evidence (Forgejo prune, CNPG, SBOM) showed evidence=0
despite successful runs. Default queries now include every
state-hub-progress event type the definitions' report/evidence sinks declare.
Three natural prune fires (2026-09-06/13/20) supply the last missing
bounded-operation evidence, so ACTIVITY-WP-0035-T08 and the workplan finish.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 241500@bnt-lap001
Assistant-Session: 4a77db80-b523-4a03-83c4-e3da08755c30
ACTIVITY-WP-0032-T02 / ACT-ADR-006. ops_runs grows harness_profile_ref (text,
indexed) and execution_refs (jsonb), migration 0008, threaded through the
emission path, queue projection, and run artefacts.
The important part is the enforcement, not the columns. ACT-ADR-006 says
approach_hint must never override, synthesize, or fall back from an absent or
invalid harness_profile_ref — a silent fallback would reintroduce the
claim-time routing failure of 2026-08-17. resolve_execution_selector never
consults the hint: a malformed ref raises even when a hint is present, and
ACTIVITY_CORE_REQUIRE_HARNESS_PROFILE turns an absent ref into an error once
definitions have migrated.
Validation is structural only, since the glas-harness catalogue is
authoritative and must not be mirrored. Requiring the <id>@<version> pin is
worth doing locally: GlasProfiles.resolve matches an unpinned ref against every
version and refuses it as ambiguous, so the pin converts a late failure into an
emission-time error without knowing any profile id.
Migration verified on real PostgreSQL 16: upgrade, downgrade, re-upgrade, and a
legacy-shaped row still inserts and stays claimable.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
llm-connect maps every provider API error onto HTTP 502 and puts the real
cause in the body (llm_connect/server.py::_error_response: error,
provider_status). LLMConnectClient.complete called raise_for_status() and threw
that body away, so a revoked OpenRouter key was indistinguishable from a downed
gateway — four days of production evidence read as "llm-connect is down".
Live check confirms one fault, not two: the llm-connect pod is Running 1/1 with
healthy endpoints, and today's FI/Binky/triage runs still 502 after yesterday's
rollout, matching the sanitized upstream 401 railiance-platform reported.
The client now raises with error, provider_status, provider, model and a
bounded copy of llm-connect's already-sanitized message, under a field
allowlist so no provider blob or key material reaches the run artefact.
Refs ACTIVITY-WP-0031-T01, T03.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Add the sbom-nexus catch_up resolver contract and the daily replacement for
weekly-sbom-staleness, both disabled until CUST-WP-0062-T03 lands.
The weekly check used `for_each: context.repos.repos` and emitted one task per
stale repo — 75 tasks on 2026-08-17 against 111/111 stale repos. The
replacement asks sbom-nexus for only the N oldest-stale repos in one ranked
call and carries no rule block at all, so tasks_spawned is 0 by construction.
- context_resolvers/sbom_nexus.py: source type `sbom-nexus`, query `catch_up`,
GET /sbom/catch-up?limit=N. Read-only; ingest is T02. Limit bounded 1..25 and
the response truncated to it so an over-long reply cannot widen T02's
side-effect.
- activity-definitions/daily-sbom-catchup.md: weekdays 09:15 Berlin, enabled:
false, deterministic sbom_catchup progress sink.
- rules/executor.py: the deterministic report builder only special-cased
context.repos, which would have emitted a contentless progress event for this
definition. _sbom_catchup_report names the selected repos and reads
updated/skipped from context when T02 populates them.
- 17 tests against a test double; no live nexus exists yet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Map every State Hub/core-hub caller to a post-retirement owner. Keep the
15-minute sweep schedule here and point the engine at repo-manager
(State Hub dual-run by default, REPO_MANAGER_URL when present). Publish
GET /execution/semantics and 410 workplan launch routes so State Hub
/execution/* is not re-homed as a task database. T03 still waits on
HUB-WP-0004.
Bare make lists targets. make install-cli installs the activity tool via uv.
Named activity-core backends live in ~/.config/activity/services.json with
list/add/use/default/which; -s/--service selects one call without changing
the default; --activity-url is a one-shot override.
Ship the activity console script for consumer-repo morning review: list,
status, runs, deliverables, inbox, checkpoint, and ack. Offline-first with
git + local defs; enriches from ops API and State Hub. Multi-source trust
matrix never reports did-not-run when git has the artefact. Adds
target_repo filter on GET /ops/automations.
ACTIVITY-WP-0027: document durable ClusterIP host access for llm-connect;
join activity_runs to ops_runs; surface Forgejo artefact links on
/ops/ui automation detail and new run detail pages.
