Repository navigation
fix(triage): chain rejected fixes straight into retriage - #60
Merged
Merged
Conversation
Previously, when FixVerifier classified a candidate fix as rejected, the workflow just swapped the issue to the `triage: fix rejected` label and stopped silently. Since that label is retriageable, the system would only pick up the rejection's feedback on a *second*, separate comment on the issue -- the comment that triggered the rejection was dropped on the floor even though it usually already contains the actionable regression info needed to retriage. Now the rejected branch posts an acknowledgment comment and chains straight into retriage() with the same conversation, so a single comment pointing out a regression in a fix branch is enough to kick off a new triage attempt.
This was referenced Nov 3, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
When
FixVerifierclassifies a candidate fix asrejected, the workflow now posts an acknowledgment comment and chains straight intoretriage()with the same conversation, instead of swapping the label totriage: fix rejectedand stopping.Why
fixRejectedis a retriageable label, so the old behavior silently parked the issue waiting for a second, separate human comment beforeRetriageJudgewould ever run — even though the comment that caused the rejection almost always already contains the actionable info needed to decide whether to retriage (as seen on withastro/astro#18600, where a maintainer's regression report + suggested fix sat idle after being classified as a rejection).How
credentialsintoverifyFix(previously only needed bytriage/retriage).rejectedverdict: swap the label, post a short acknowledgment comment with the verdict's reasoning, then callthis.retriage(...)with the issue's current label set tofixRejectedsoRetriageJudgeevaluates the same conversation immediately.Testing
tsc --noEmitclean.vitest run: 425/427 pass; the 2 failing tests (tests/ai-gateway.test.ts) are pre-existing and unrelated (confirmed failing onmainbefore this change too — stale model-catalog fixture).TriageWorkflow.runto extend (coverage here is viatriage-fsm.test.ts/triage-labels.test.tsunit tests, which are unaffected).