You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
source-control:pull-request: bare 'Refs #N' opt-out clears the skill's local gate but not the real pr-issue-linkage validator #630
Lane: work · Triggering example: discovered while implementing #487 (PR #629, source-control 0.12.0 → 0.12.1).
Observed vs expected
plugins/source-control/skills/pull-request/reference/create.md §2.4.2 treats a bare Refs #N line as a valid closing-keyword opt-out for its own local gate. The real pr-issue-linkage reusable CI workflow (pinned ci-workflows@90f1c54 at time of writing) only accepts a genuine native GitHub closing keyword (Closes/Fixes/Resolves + #N) or one of the documented opt-out markers (No linked issue, No related issue:) — Refs #N is not in that set.
Observed: a PR body that satisfies the skill's own §2.4.2 opt-out check can still fail the real pr-issue-linkage CI gate, because the two acceptance sets diverge. Expected: the skill's local opt-out marker set and the CI workflow's validator regex should accept the identical set of markers, so a PR the skill considers "gate-satisfied" always is.
Category
behavioral (contract mismatch between a skill's self-check and the CI gate it's meant to satisfy)
Suggested fix direction (not investigated in depth — pick one)
Drop Refs #N from the skill's accepted opt-out set, matching the validator exactly.
Extend the validator's accepted marker set to include Refs #N (requires a ci-workflows change per this org's cross-repo ownership — Actions-execution behavior lives upstream there).
Option 1 is the narrower, same-repo fix; option 2 changes shared CI behavior for every consumer.
Related
#487 (source, closes) · PR #629 (where this was found and documented in the PR body / create.md §2.4.1 prose as a known asymmetry, not fixed in that PR's scope)
Single clear fork with the issue's own recommendation already load-bearing: option 1 (drop Refs #N from the skill's local opt-out set, matching the real pr-issue-linkage validator) is same-repo and narrow; option 2 (extend the validator to accept Refs #N) requires a cross-repo ci-workflows change affecting every consumer of that reusable workflow — out of this repo's ownership per this org's cross-repo convention, and a much larger blast radius for a self-inflicted mismatch.
Decision defaulted: option 1 — veto before merge. Drop Refs #N from create.md §2.4.2's accepted opt-out set so the skill's local gate matches the real CI validator exactly.
Activity
kyle-sexton commentedon Jul 20, 2026
This was generated by AI during triage.
Triage — ready with noted decision.
Single clear fork with the issue's own recommendation already load-bearing: option 1 (drop
Refs #Nfrom the skill's local opt-out set, matching the realpr-issue-linkagevalidator) is same-repo and narrow; option 2 (extend the validator to acceptRefs #N) requires a cross-repoci-workflowschange affecting every consumer of that reusable workflow — out of this repo's ownership per this org's cross-repo convention, and a much larger blast radius for a self-inflicted mismatch.Decision defaulted: option 1 — veto before merge. Drop
Refs #Nfromcreate.md§2.4.2's accepted opt-out set so the skill's local gate matches the real CI validator exactly.fix(source-control): route multi-issue Refs to Related and align orph…
fix(source-control): accept Refs: markers and block negated closers i…