[detection-analysis] Detection Analysis Report — 2026-10-30 (last 24h)

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC), 2026-10-29T23:14Z → 2026-10-30T23:12Z
• Total runs analyzed: 490 (executed runs with full metadata); an additional 185 runs surfaced only minimal GitHub Actions metadata (no aw_info.json/run_summary.json) and were excluded from detection classification — coverage of those runs is incomplete
• Detection-enabled runs: 471 (96.1%)
• Regular runs (gh-aw-detection: false): 19 (3.9%)
• Misconfigured workflows found: 0

Note

No gh-aw-detection-related misconfigurations were found in this window. All 19 regular (detection-disabled) runs come from one-off "Smoke ..." test workflows with a single run each — none crossed the >3-runs-in-the-window threshold used as a proxy for rule 1, and no detection job itself failed (rule 2: 448/454 detection jobs present succeeded, 6 skipped, 0 failed). No workflow alternated between enabled/disabled (rule 3).

Comparison Chart

Metric	Regular Runs	Detection Runs
Run count	19	471
Success rate	84.2%	67.7%
Failure count	3	152
Avg token usage	105,571	170,305

The lower success rate on detection-enabled runs is not evidence the feature itself causes failures — the "regular" bucket is entirely single-shot smoke tests (simple, low-token, high-pass-rate by design), while the detection-enabled bucket is the full population of production/scheduled workflows with naturally more complex tasks and higher token usage. No detection-job failure was observed anywhere in the data (see Misconfigured Workflows).

Misconfigured Workflows

No misconfigured workflows detected in this window.

View All Run Metrics

Workflows with more than one run in the window (18 workflows, 331 of 490 classified runs; the remaining 159 workflows each had exactly 1 run, mostly one-off "Smoke ..." engine tests):

Workflow	Total Runs	Detection	Regular	Success Rate	Avg Tokens
PR Sous Chef	114	114	0	0.0%	5,587
Issue Monster	47	47	0	91.5%	79,625
Matt Pocock Skills Reviewer	21	21	0	95.2%	44,482
Test Quality Sentinel	21	21	0	95.2%	9,680
Design Decision Gate 🏗️	20	20	0	90.0%	2,496
PR Code Quality Reviewer	20	20	0	100.0%	27,275
Ponytail Reviewer	20	20	0	100.0%	185,428
Avenger	16	16	0	100.0%	669,562
Daily Go Test Parallelizer	12	12	0	100.0%	1,146,434
AI Moderator	10	10	0	90.0%	1,042
Auto-Triage Issues	6	6	0	100.0%	131,763
Contribution Check	6	6	0	100.0%	18,344
[aw] Failure Investigator (6h)	4	4	0	100.0%	477
PR Triage Agent	4	4	0	100.0%	88,467
Code Scanning Fixer	4	4	0	100.0%	11,002
Agent Job Health Monitor	2	2	0	0.0%	122,025
Claude Code User Documentation Review	2	2	0	100.0%	69,044
Daily Credit Limit Test	2	2	0	100.0%	46,394

All 19 explicitly-disabled (gh-aw-detection: false) runs belong to distinct single-run "Smoke ..." engine-validation workflows (e.g. Smoke Claude, Smoke Codex, Smoke Pi, Smoke Copilot and its variants, Smoke Agy, Smoke OTEL) — a deliberate, low-volume configuration, not a misconfiguration.

View Historical Trend

31 days of history are available (2026-08-25 → 2026-10-30, 32 recorded data points):

Recommendations

1. No detection-feature action required — the gh-aw-detection flag is configured correctly across all observed workflows this window; keep monitoring as new workflows are added.
2. Out of scope but notable: PR Sous Chef failed 114/114 runs in this window. The detection job succeeded every time and the agent job itself mostly succeeded (112/114) — the failure boundary is the safe_outputs / "Process Safe Outputs" step, i.e. an agent-logic/safe-outputs issue, not a detection misconfiguration. Worth a separate investigation (example: §40972856581).
3. Coverage gap: 185 of 675 discovered runs in the window only had minimal GitHub Actions metadata (no aw_info.json), so their detection status is unknown and they were excluded from these metrics. If precise 100% coverage is required, re-run collection with a higher max_storage/max_github_api_rate_limit budget.
4. Rule 1 ("gh-aw-detection: false with >3 runs") was evaluated against this 24h window as a proxy for the intended 7-day lookback; re-validate over a full 7-day window before drawing long-term conclusions about any workflow's opt-out status.

References:

• §40972856581 — PR Sous Chef (0% success, out-of-scope safe_outputs failure)
• §41004654784 — Daily GitHub Docs SEO Optimizer (example regular/detection-disabled run)
• §40971733897 — Agent Job Health Monitor (example detection-enabled run with a failure, detection job itself succeeded)

Generated by 🔎 Detection Analysis Report · claude · agent · ~109 AIC · ⌖ ~1 AIC · ⊞ 4.4K · ◷

• expires on Nov 2, 2026, 3:29 PM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Detection Analysis Report.

A newer discussion is available at Discussion #73579.