[regression] Daily CI Regression Audit — 31765617517

[github-actions]

Summary

• CI failure rate (last 20 runs of ci.yml): 9 failures / 20 runs = 45% (plus 5 cancelled, 6 success)
• Consecutive failure streak: up to 3 in a row (runs #53234–#53236)
• Recommended action: Investigate — failure rate is well above the 10% threshold and includes a notable streak.

Failure Details

Top failure patterns (from 3 most recent failed runs):

1. Integration test package failures with no individual test flagged as failed — pkg/workflow (run #53241, job "Integration: Workflow Infra") and pkg/cli (run #53236, job "Integration: CLI Completion & Other") both ended with FAIL at the package level but the harness logged "⚠️ No individual test marked as failed!", suggesting a build/init issue or race condition rather than a simple assertion failure.
2. TestCacheMemoryMultipleIntegration failing intermittently in pkg/workflow (subtest single_cache-memory_(backward_compatible)), seen in run #53241.
3. TestInputValidation failing in pkg/cli (subtests around key=value parsing: invalid_input_without_equals, invalid_input_-_empty_key, mixed_valid_and_invalid), seen in run #53236.

These look like flaky/non-deterministic integration test behavior (race conditions or shared test state) rather than a single clear-cut regression, since the same suites intermittently pass in other runs.

Open regression/bug-labeled issues (labels: bug, regression, top 5):

• [aw-failures] [P0] Crush CLI ENOENT in sandboxed exec — Daily Code Metrics failing 2 days running #52571 — [aw-failures] [P0] Crush CLI ENOENT in sandboxed exec — Daily Code Metrics failing 2 days running
• [aw-failures] [aw] Failure Investigator Report — 2026-08-13 19:10 UTC (6h) #52570 — [aw-failures] [aw] Failure Investigator Report — 2026-08-13 19:10 UTC (6h)
• [aw-failures] [P1] PR Sous Chef: Start DIFC Proxy step fails with no captured root cause #52502 — [aw-failures] [P1] PR Sous Chef: Start DIFC Proxy step fails with no captured root cause
• [aw-failures] [P0] Failure-Investigator log capture grabs post-job cleanup tail, not the actual error #52501 — [aw-failures] [P0] Failure-Investigator log capture grabs post-job cleanup tail, not the actual error
• [aw-failures] [P1] Go Logger Enhancement fires 6 conflicting safe-output tools per retry attempt #52460 — [aw-failures] [P1] Go Logger Enhancement fires 6 conflicting safe-output tools per retry attempt

None of these open issues directly cover the pkg/workflow/pkg/cli integration test flakiness identified above — this may warrant a new tracking issue.

Generated by 🧭 Daily Regression Audit · auto · 52 AIC · ⌖ 2.41 AIC · ⊞ 6.4K · ◷

• expires on Aug 15, 2026, 7:10 PM UTC-08:00