[aw-failures] [aw-fix] Avenger (Claude engine) hard-fails at startup — "no structured log entries were produced" (ERR_CONFIG) be [Content truncated due to length] #44303

Fix Avenger's Claude-engine startup path — it produced zero structured log entries and hard-failed ERR_CONFIG on 3 runs in the last 6h, blocking the whole workflow before any tool executes.

The Claude engine exits before emitting a single structured log entry, so parse_claude_log.cjs has nothing to parse and the agent job fails at the Parse agent logs for step summary step. This is a startup/configuration failure, not a tool-execution failure.

Failure signature

##[error]ERR_CONFIG: Claude execution failed: no structured log entries were produced.
This usually indicates a startup or configuration error before tool execution.

Affected workflow & runs (last 6h, github/gh-aw)

Engine: claude (OTEL_RESOURCE_ATTRIBUTES=...gh-aw.engine.id=claude). 3 Avenger failures total in the window; the representative run is confirmed to carry the ERR_CONFIG signature.

Probable root cause

The Claude CLI process terminated during startup/config generation before writing any JSONL structured-log line. Likely candidates, in order:

1. Transient inference/auth availability at Claude startup (rate-limit, 4xx, or empty response) that aborts before the first log line.
2. A configuration/setup error surfaced only as a non-zero exit with empty stdout — the current handler collapses all such cases into a single opaque ERR_CONFIG.

The symptom handler (parse_claude_log.cjs → ERR_CONFIG) is correct to fail, but it discards the underlying cause, making the failure unactionable from logs alone.

Proposed remediation

1. Surface the real startup error. Before raising ERR_CONFIG, capture and echo the Claude CLI's raw stderr/exit code into the step summary so "no structured log entries" points at the actual cause (auth, rate-limit, config) instead of a generic message.
2. Add a bounded startup retry. If the engine exits with zero structured entries and a retriable signal (empty output / transient 429/5xx), retry the Execute step 1–2 times with backoff before failing the job.
3. Distinguish transient from config. Emit a specific output flag (e.g. inference_access_error / ai_credits_rate_limit_error, which already exist as job outputs) when the empty-log exit is attributable to inference availability, so transient outages aren't reported as configuration bugs.

Success criteria / verification

• A run whose Claude engine exits before producing structured logs shows the underlying stderr/exit reason in the step summary (no bare ERR_CONFIG).
• Transient empty-startup exits are retried and, when the retry succeeds, the workflow completes green.
• Genuinely transient inference outages are classified via the existing inference_access_error / ai_credits_rate_limit_error outputs rather than ERR_CONFIG.
• Over a subsequent 24h window, Avenger no longer accumulates repeated ERR_CONFIG-with-empty-logs failures.

Notes

• Not a duplicate: existing open trackers cover different signatures — [aw-failures] [aw-fix] Smoke CI sandbox setup hard-fails — stale root-owned /tmp/gh-aw/sandbox/firewall blocks AWF (uid=1001) #44242 (Smoke CI firewall EACCES), [aw-failures] [aw-fix] Impeccable Skills Reviewer times out at 15m after completing its review — Copilot CLI hangs on exit #44243 (Impeccable Skills Reviewer Copilot CLI hang/timeout), [aw-failures] Copilot BYOK 100% red — SDK driver sends OpenAI stream_options to type=anthropic provider, upstream 400 kills [Content truncated due to length] #43031 (Copilot BYOK stream_options). None address the Claude-engine empty-startup path.
• The dominant failure cluster this window (Smoke CI EACCES /tmp/gh-aw/sandbox/firewall/logs, uid=1001) is already tracked by [aw-failures] [aw-fix] Smoke CI sandbox setup hard-fails — stale root-owned /tmp/gh-aw/sandbox/firewall blocks AWF (uid=1001) #44242 and remained active as recently as run §28944424412 (12:58 UTC) — left open, no new issue.

References:

• §28944237120
• §28937890974
• §28934652958

Generated by 🔍 [aw] Failure Investigator (6h) · 152 AIC · ⌖ 14.3 AIC · ⊞ 6.2K · ◷

• expires on Jul 15, 2026, 5:29 AM UTC-08:00