`ADHOC-YYYY-MM-DD` is unique per date but not per repository, so any two repos
opening an ad-hoc on the same day collide. The 2026-08-26 fleet projection
reset refused 9 records for exactly this reason.
Canon (work-record-types_v0.1, CUST-WP-0066) settled the form as
`{PREFIX}-WP-ADHOC-YYYY-MM-DD`, filename unchanged, and grandfathered existing
ids on the condition they are never *silently* re-derived. This is the explicit
migration that clause allows for.
The hub id is derived from the record id, so a changed id is a different
record: stale state_hub_*_id fields are dropped and fix-consistency re-derives.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 2583210@bnt-lap001
Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
Canon defines six work-record kinds — workplan, task, intake, decision,
engagement, register-entry. These files declared none of them (feature, runbook,
bug-report, extension-workplan, scorecard, session-note, adhoc) while the hub
held each as a workplan.
Selection is by 'type: workplan', so the mismatch made them invisible to any
projection: a forge-derived reset read their correct hub records as no longer
deriving and proposed retiring live work.
Where a file also lacked a canonical identifier, one was assigned continuing the
repository's sequence. Existing hub identifiers are preserved, so no record is
orphaned.
Refs STATE-WP-0083
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 2583210@bnt-lap001
Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
Updated by fix-consistency on 2026-08-25:
- update .custodian-brief.md for activity-core
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 2583210@bnt-lap001
Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
These workplans exist only in the retired local hub. Their random pre-ADR-007
identifiers are refused by C-06 as stale references, so they cannot be
registered. Deriving from the canonical record id takes no identity from
anything: central does not hold them and the old ids die with the cache.
Records central already holds were deliberately left untouched.
Refs CUST-WP-0068-T06
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 2583210@bnt-lap001
Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
Updated by fix-consistency on 2026-08-25:
- update .custodian-brief.md for activity-core
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 2583210@bnt-lap001
Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
The remote row pointed at 127.0.0.1:18000, a reverse tunnel back to the
workstation. On railiance01 the State Hub runs in the cluster on that same
machine, so the request left the box and came back to reach a local service.
Refs CUST-WP-0067-T07
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 2583210@bnt-lap001
Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
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>
glas-harness answered both open questions from ACTIVITY-WP-0032, so T01 is
resolved as option A: the ops_run pull queue stays and carries the versioned
harness_profile_ref plus attribution refs. The claiming executor passes the
request into Glas, which resolves or refuses before sandbox creation. This
changes the execution contract without also changing scheduling topology.
Two answers shaped the plan:
- harness_profile_ref and approach_hint coexist with distinct semantics.
approach_hint must never override, synthesize, or fall back from an absent or
invalid profile ref on governed execution — T02 must enforce that in code.
- The glas-harness catalog is authoritative and must not be mirrored. Glas has
no network validation service today, so emit-time remote validation is not
available.
T03 is corrected accordingly: validate locally and structurally, rely on the
execution-side Glas resolver as the mandatory fail-closed check, and record the
residual gap rather than papering over it with a mirrored catalogue.
Hub decision 147beec6-7fe7-4837-8e3a-4264a240379d.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Tested rather than relayed: OpenRouter answers 'User not found' for the
delivered key, which means no account resolves it — an existing-account key
reissue, not a new account. Ruled out credits (402), model permissions (403),
provider config, and pod health.
Also records a latent delivery defect: the secret carries a trailing newline,
harmless today only because llm_connect/config.py:53 strips it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
scripts/prod_automation_status.sh took a positional since value, so the
Makefile's own `SINCE=sunday` form passed the literal string into timestamptz:
all five sections errored while the run still looked like it produced a report.
Now `SINCE=` is accepted and the value is parsed and normalised to UTC up
front, failing fast with usage and exit 2 instead of five parse errors.
ACTIVITY-WP-0032 plans adoption of the glas-harness contract 1.0 reported in
GLAS-WP-0004. WP-0026 is finished, so this gets its own plan. The motivation is
concrete: ops_run.approach_hint binds at claim time, which produced a failed
run on 2026-08-17 ("no approach matched labels/definition") after it had
already consumed a claim and a lease.
T01 is deliberately blocking: our claim path is pull-based and the Glas
contract is a call, so the invocation shape is an architectural decision, not a
port. Two questions are outstanding with glas-harness.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
