[ci-scan] Tighten agentic CI workflow dedup and feedback emission#129231
Merged
kotlarmilos merged 2 commits intoJun 11, 2026
Conversation
Detection (ci-failure-scan + shared KBE instructions): the existing-KBE search was open-only, so a closed `[ci-scan]` predecessor or an unlabeled human "Test failure: ..." report was invisible and the same failure got re-filed across runs (e.g. X509Certificate2CollectionRemoveRangeArray, crossgen2_comparison.py). Add a no-label open-issue search variation and a recently-closed-KBE search that records `skipped: recently-closed dup #<n>` unless the signature genuinely recurs after the predecessor was closed. Feedback (ci-failure-scan-feedback): on a scheduled run there is no triggering PR, so `push_to_pull_request_branch`/`update_pull_request` must carry the found PR's `pull_request_number`. Also emit the KPI tracker `update_issue` before the PR safe-outputs, so a push or patch failure can no longer cancel the daily tracker snapshot. Prompts are runtime-imported, so no lock regeneration is required. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Contributor
There was a problem hiding this comment.
Pull request overview
This PR updates the agentic CI workflow prompts/instructions to improve (1) duplicate detection when filing Known Build Errors (KBEs) and (2) robustness of feedback/KPI emission in scheduled runs.
Changes:
- Extend the shared KBE dedup search to also catch unlabeled human-filed issues and to consider recently-closed KBEs.
- Update the CI failure scan workflow to follow the new “recently-closed” dedup section and record the new skip outcome.
- Adjust the feedback workflow instructions to pass
pull_request_numberon scheduled PR updates and to emit the KPI tracker output before PR safe-outputs.
Show a summary per file
| File | Description |
|---|---|
| .github/workflows/shared/create-kbe.instructions.md | Adds a “recently-closed KBEs” search section and an open-issue no-label search variation to reduce duplicate KBE creation. |
| .github/workflows/ci-failure-scan.md | Updates the scan workflow to follow the new dedup section and to recognize the new “recently-closed” skip outcome. |
| .github/workflows/ci-failure-scan-feedback.md | Updates feedback emission guidance for scheduled runs (explicit PR number) and orders KPI tracker emission ahead of PR safe-outputs. |
Copilot's findings
- Files reviewed: 3/3 changed files
- Comments generated: 3
kotlarmilos
changed the title
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
vitek-karas
approved these changes
Jun 11, 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.
Description
The detection step only searched open issues, so it could miss a recently-closed predecessor or an unlabeled human report of the same failure and file a duplicate Known Build Error; this change adds a no-label open-issue search and a recently-closed search so a repeat failure is skipped for human review instead of refiled, unless it genuinely reproduces after the earlier issue was closed, while the feedback step now passes the pull request number explicitly when it pushes or updates a branch on a scheduled run (which previously failed because there is no triggering pull request) and emits the KPI tracker update before the pull request outputs so a push or patch failure can no longer drop the daily snapshot, and because the prompts are imported at runtime no compiled workflow files need regenerating.