mirror of
https://github.com/QwenLM/qwen-code.git
synced 2026-08-09 08:45:58 +00:00
d4c15f05c5
* feat(ci): add stale failure patrol * fix(ci): harden failure patrol * refactor(ci): simplify flaky rerun patrol * docs(ci): clarify flaky patrol skill boundary * feat(ci): patrol stale PR failures * fix(ci): prefilter failed PRs * fix(ci): isolate patrol classification * fix(ci): revalidate stale patrol actions * fix(ci): verify main before branch update * fix(ci): classify all stale PR failures * fix(ci): bound patrol batches * fix(ci): persist patrol failure state * fix(ci): harden patrol state transitions * fix(ci): harden stale failure patrol * fix(ci): continue patrol after expired logs * fix(ci): tighten patrol guardrails * fix(ci): preserve failure context in patrol logs * fix(ci): harden stale patrol closeout * fix(ci): paginate patrol marker comments * fix(ci): harden patrol action guards * test(ci): cover patrol guard rails * fix(ci): harden patrol marker parsing * fix(ci): address patrol review followups * test(ci): cover patrol review edges * fix(ci): record patrol rerun marker first * fix(ci): harden patrol review edge cases * fix(ci): tighten stale failure patrol markers * refactor(ci): simplify flaky rerun patrol (2838→1258 lines) - Remove classification guards from actOnDecision (confidence check, action enum validation, boundedReason, update_branch multi-guard chain) - Move classification rules to SKILL.md prompt - Delete 40 source-code text matching tests, keep 20 behavior tests - Merge identity job into classify, remove SHA verification - Change scan sort order from oldest-first to newest-first - Remove unused functions: writeSkillInputs, failureKey, boundedReason, canAct, skillCandidate, mainRunSucceeded * fix(ci): show gh stderr in top-level error output When gh CLI returns non-zero exit, execFile rejects with an error whose .stderr contains the actual GitHub API diagnostic. Previously only one of stderr or message was shown; now both are printed. * fix(ci): address patrol review findings * fix(ci): make stale patrol actions recoverable * refactor(ci): simplify flaky rerun patrol * fix(ci): close flaky patrol review gaps * fix(ci): restore PR failure patrol actions * fix(ci): harden failure patrol scanning * fix(ci): address patrol review follow-ups * fix(ci): harden patrol parsing and coverage * fix(ci): classify failures against PR changes * fix(ci): harden patrol script input handling * fix(ci): remove unsafe auto branch update * test(ci): exercise patrol action limit * fix(ci): bind patrol actions to current evidence * fix(ci): count patrol actions per PR * fix(ci): redact quoted secret labels --------- Co-authored-by: Shaojin Wen <shaojin.wensj@alibaba-inc.com> |
||
|---|---|---|
| .. | ||
| audio-capture-prebuilds.yml | ||
| build-and-publish-image.yml | ||
| cd-cua-driver.yml | ||
| cd-mobile-mcp.yml | ||
| check-issue-completeness.yml | ||
| ci.yml | ||
| codeql.yml | ||
| comment-attachment-guard.yml | ||
| desktop-release.yml | ||
| docs-page-action.yml | ||
| e2e.yml | ||
| finalize-release.yml | ||
| gemini-scheduled-pr-triage.yml | ||
| main-ci-failure-issue.yml | ||
| pr-force-push-reminder.yml | ||
| qwen-autofix.yml | ||
| qwen-automated-issue-triage.yml | ||
| qwen-ci-flaky-rerun.yml | ||
| qwen-code-pr-review.yml | ||
| qwen-issue-followup-bot.yml | ||
| qwen-pr-safety-precheck.yml | ||
| qwen-scheduled-issue-triage.yml | ||
| qwen-triage.yml | ||
| release-sdk-python.yml | ||
| release-sdk.yml | ||
| release-vscode-companion.yml | ||
| release.yml | ||
| sdk-python.yml | ||
| stale.yml | ||
| sync-cua-driver-to-oss.yml | ||
| sync-release-to-oss.yml | ||
| terminal-bench.yml | ||