[aw-failures] [P1] PR Description Updater: update_pull_request rejects genuine pull_request-triggered runs with "empty event" er [Content truncated due to length]

[github-actions]

Problem

Fix update_pull_request's target: triggering resolution — it rejects a genuine pull_request: closed (merged) trigger as if no PR context exists, so every invocation of PR Description Updater that actually has content to write fails. This isn't an edge case: it's the workflow's only real code path.

Affected workflows and runs

• PR Description Updater (.github/workflows/pr-description-caveman.lock.yml) — §32051724487 (2026-08-17 17:45 UTC). Triggered by a genuine pull_request event (PR [test-parallel] test: add t.Parallel() to safe subtest in audit_mcp_tool_usage_test.go #53438, merged into main, not stacked, not cross-repo).

Probable root cause

The safe_outputs job's update_pull_request handler isn't seeing the real event context. The agent's tool call failed twice with: "update_pull_request requires a pull request context but the workflow is running on a \"\" event. target: triggering only applies when a pull request triggered the workflow." — first without pull_request_number, then again with pull_request_number: 53438 explicitly. Both attempts hit the identical error.

But gh run view 32051724487 confirms event: pull_request for this run, and the same run's agent job resolves github.event.pull_request.number correctly elsewhere (used for diff fetching). So the event context exists at the GitHub Actions level — the break is specifically inside the safe-outputs handler's own target: triggering resolution, which appears to read event name/PR number independently and gets an empty result even on a legitimately PR-triggered run.

Proposed remediation

1. Trace why target: triggering evaluates event_name as empty inside the safe_outputs job container when the triggering GitHub Actions event genuinely was pull_request.
2. Add a regression test: a workflow with update-pull-request: {operation: replace, max: 1} (no explicit target) triggered by pull_request: closed must resolve the PR number without requiring target: '*' plus an explicit pull_request_number.
3. audit-diff against the nearest earlier run of the same workflow (§32044038495) found no usable comparator — that run took the noop path (0 GitHub API calls, no diff content) and never exercised update_pull_request. Can't yet confirm this is a new regression vs. a standing bug; treat it as a persistent break until a prior successful update_pull_request invocation on this workflow is found.

Success criteria / verification

• A pull_request: closed (merged)-triggered run of PR Description Updater with real diff content successfully calls update_pull_request and updates the PR body — no target: triggering rejection.
• Confirm on a fresh merged PR through this workflow (or a targeted replay of run 32051724487's scenario).

Analyzed run IDs: 32051724487 (representative), 32044038495 (comparator — inconclusive, no update_pull_request invocation to diff against).

Parent: #53129
Related to #53129

Generated by 🔍 [aw] Failure Investigator (6h) · agent · 157.7 AIC · ⌖ 9.12 AIC · ⊞ 5.8K · ◷

• expires on Aug 24, 2026, 11:12 AM UTC-08:00

[github-actions]

🍪 Issue Monster selected this for Copilot

I've identified this issue as a good candidate for automated resolution and requested assignment to the Copilot coding agent.

If assignment succeeds, the Copilot coding agent will analyze the issue and create a pull request with the fix.

Om nom nom! 🍪

🍪 Om nom nom by Issue Monster · gpt54 · 2.67 AIC · ⌖ 6.67 AIC · ⊞ 10.6K · ◷