Pause daily-todo-md-stale-review: sole source of the 5 closed Forgejo issues
Traced the origin of the-custodian issues #1-5 (opened 2026-07-08, all reviewed and closed 2026-07-20): this ActivityDefinition's matched rule always produces a TaskSpec (rules/actions.py::_task_spec_for_rule), and RunActivityWorkflow unconditionally routes TaskSpecs through emit_tasks -> the deployment-wide IssueSink (ISSUE_SINK_TYPE=rest -> issue-core -> Forgejo). It is the fleet's only consumer of the task_template rule action -- confirmed via grep across activity-definitions/. Per current policy (Forgejo issue tracking is not the fleet coordination mechanism), paused rather than left to recreate the same sidetrack daily: enabled: false, status: paused, re-enable path documented in the file (needs a new activity-core report-sink authoring a kind: intake work record instead of an IssueSink task spec). Added as item 6 to CUST-WP-0060's stage-3 successor seed, including the reminder to grep the fleet for any other task_template consumer before stage 3 ships. This is the ADR-001 file-level fix; the live activity-core DB row and Temporal schedule pick it up on the next Railiance-deployed 'make sync-activity-definitions' run -- not something I have credentials or standing to trigger directly from this session. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
parent
36f0336ca6
commit
90477f3f0d
2 changed files with 42 additions and 3 deletions
|
|
@ -181,3 +181,15 @@ with green proof runs.
|
|||
generate *around* them.
|
||||
5. Suggestion-table close-out (read-only legacy) once the intake entity is
|
||||
live.
|
||||
6. **New report-sink for activity-core** (added 2026-07-20, Forgejo-issue
|
||||
housekeeping session): `daily-todo-md-stale-review` was found still
|
||||
routing through the deployment-wide `IssueSink` (`ISSUE_SINK_TYPE=rest`
|
||||
→ issue-core → Forgejo) — the only fleet consumer of the `task_template`
|
||||
rule action, and the source of 5 stale-review issues nobody was meant
|
||||
to see (Forgejo issue tracking is not the fleet coordination mechanism).
|
||||
Paused (`enabled: false`) rather than left running; re-enable requires
|
||||
a new activity-core report-sink type that authors a `kind: intake` work
|
||||
record in the target repo (per this workplan's stage-3 promotion
|
||||
design) instead of an `IssueSink` task spec. Any other `ActivityDefinition`
|
||||
that later adds a `rule.action.task_template` inherits the same trap —
|
||||
worth a fleet-wide grep before stage 3 ships.
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue