[aw-failures] GitHub Copilot CLI generic exit-1 crash across 7 workflows

[github-actions]

Fix the crash first, then add exit-1 diagnostics — the Copilot CLI engine is exiting mid-run with zero error detail, and it's already spread to 7 unrelated workflows in one 5-hour window.

Problem statement

The native GitHub Copilot CLI engine (engine: copilot, distinct from the Codex/Claude "Copilot-brokered credential" bug tracked in #55412/#55413) is crashing mid-run with a bare ##[error]Agent execution exited with code 1 and no further diagnostic. Firewall logs are clean (no blocked requests, no auth-proxy involvement), and the crash happens after real tool activity — this is not a startup/auth failure, it's a genuine in-flight crash.

Affected workflows and run IDs

• Linter Miner — §32757223241 (audited: 82/82 firewall requests allowed, 22 bash + 3 task calls completed before crash)
• Daily Copilot PR Merged Report — §32744281191
• Delight — §32741761003
• Architecture Guardian — §32738225987
• Daily Malicious Code Scan Agent — §32733970151
• CLI Consistency Checker — §32732771271
• Code Scanning Fixer — §32731261007

Probable root cause

Unknown and currently unlogged, same shape as the previously-seen "generic exit 1, no captured stdout/stderr reason" pattern noted in #54242 for Codex — except this is the Copilot CLI engine, a different binary/wrapper. The clean firewall trace and mid-run tool activity rule out auth/proxy and network causes. Two live hypotheses: (1) the Copilot CLI binary itself is crashing/OOMing on larger tool-call sequences, or (2) the harness's exit-code capture for this engine is swallowing the real stderr before it reaches the step summary — worth checking whether the wrapper script for engine: copilot captures stderr on non-zero exit the same way the Codex/Claude wrappers do.

Proposed remediation

1. Add stderr/exit-reason capture to the Copilot CLI engine wrapper so future crashes surface something more specific than "exited with code 1" (compare to how [aw-failures] [P0] Cross-engine agent-CLI segfault (exit 139) — confirmed on Auto-Triage/Pi CLI, now spread to 8 workflows acros #54186 escalated once exit code 139 was confirmed instead of "generic exit code 1").
2. Re-run one of the affected workflows (e.g. Linter Miner) with verbose/debug logging enabled for the Copilot CLI step to capture the actual crash reason.
3. Check for a resource ceiling (memory/timeout) being hit after ~20+ tool calls, since Linter Miner's crash occurred right after a burst of bash/task activity.

Success criteria / verification

• Next occurrence of this crash produces a specific error message (not bare exit code 1) in the step log.
• Re-run of at least one previously-failing workflow (Linter Miner or Code Scanning Fixer) completes successfully under the same trigger conditions.
• No recurrence across the 7 affected workflows over the following 24h monitoring window.

Generated by 🔍 [aw] Failure Investigator (6h) · claude · agent · 161.3 AIC · ⌖ 8.19 AIC · ⊞ 6.4K · ◷

• expires on Aug 31, 2026, 11:15 AM UTC-08:00

[github-actions]

🍪 Issue Monster selected this for Copilot

I identified this issue as a strong 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 · codex · mai10 · 2.07 AIC · ⌖ 1.36 AIC · ⊞ 18.2K · ◷