[detection-analysis] Detection Analysis Report — 2026-10-19

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC), 2026-10-18T23:21:57Z → 2026-10-19T23:21:57Z
• Total runs analyzed: 598 (full window coverage confirmed — initial download hit a 3,000-item/1 GB continuation limit, so it was paged backward through 5 fetches until the continuation cursor reached the window start with no partial flag remaining)
• Detection-enabled runs: 519 (86.8%)
• Regular runs (gh-aw-detection: false): 79 (13.2%)
• Misconfigured workflows found: 3

Warning

3 workflows explicitly disable gh-aw-detection while running more than 3 times in the window (Rule 1). All three are smoke-test workflows (Smoke Copilot, Smoke Copilot SDK Routed, Smoke Pi Routed) with the override hardcoded in their .md frontmatter — likely intentional for engine-routing smoke tests, but flagged per the configured threshold since no exception/comment documents the intent. No detection-job-level failures (Rule 2) or flapping configurations (Rule 3) were found.

Comparison Chart

Metric	Regular Runs	Detection Runs
Run count	79	519
Success rate	73.4%	81.9%
Failure count	21	94
Avg. tokens	135,940	148,432

Misconfigured Workflows

Workflow	Misconfiguration Type	Run Count	Recommended Fix
Smoke Copilot	gh-aw-detection: false on a high-frequency workflow (Rule 1)	4	Confirm the disable is intentional for engine-routing smoke testing; if so, add a frontmatter comment documenting why, otherwise remove the override.
Smoke Copilot SDK Routed	gh-aw-detection: false on a high-frequency workflow (Rule 1)	4	Same as above.
Smoke Pi Routed	gh-aw-detection: false on a high-frequency workflow (Rule 1)	4	Same as above.

Example run: §39176941475 (Smoke Copilot)

View All Run Metrics

Top 20 workflows by run count in the window:

Workflow	Runs	Detection	Regular	Success %	Avg Tokens
PR Sous Chef	76	76	0	98.7%	1,035
Issue Monster	44	44	0	86.4%	132,049
Test Quality Sentinel	32	32	0	81.2%	12,245
Ponytail Reviewer	29	29	0	86.2%	140,180
PR Code Quality Reviewer	28	28	0	50.0%	92,307
Matt Pocock Skills Reviewer	28	28	0	82.1%	48,890
Design Decision Gate 🏗️	27	27	0	85.2%	5,887
AI Moderator	17	17	0	100.0%	896
Avenger	13	13	0	92.3%	966,892
Daily Go Test Parallelizer	12	12	0	91.7%	925,917
Auto-Triage Issues	11	11	0	90.9%	112,726
Contribution Check	6	6	0	33.3%	16,102
[aw] Failure Investigator (6h)	6	6	0	83.3%	310
Smoke Copilot Routed	5	5	0	100.0%	96,205
Smoke Copilot	4	0	4	50.0%	10,785
Smoke Copilot SDK Routed	4	0	4	100.0%	2,313
Smoke Pi Routed	4	0	4	75.0%	26,635
PR Triage Agent	4	4	0	100.0%	67,764
Code Scanning Fixer	4	4	0	100.0%	10,811
Smoke Claude Dynamic Workflow	4	4	0	75.0%	1,362

236 distinct workflows ran in the window; the remainder were mostly single/low-frequency smoke-test workflows with 1–3 runs each.

View Historical Trend

9 daily snapshots over the last 30 days (cache-memory trending history). Regular-run volume has grown from single digits to 21–79/day over the past week, while detection success rate has fluctuated 76–93%; today's 81.9% sits mid-range.

Recommendations

• Document or remove the gh-aw-detection: false override in smoke-copilot.md, smoke-copilot-sdk-routed.md, and smoke-pi-routed.md — add a short frontmatter comment explaining the smoke-test rationale so future audits don't re-flag it, or re-enable detection if there's no longer a reason to skip it.
• No detection-job failures were observed this window — the detection feature itself appears healthy across the 519 detection-enabled runs.
• Regular-run success rate (73.4%) continues to trail detection-run success rate (81.9%); worth a follow-up to confirm this gap is driven by the smoke-test workloads above rather than a broader reliability issue in non-detection workflows.

References:

• §39176941475 — Smoke Copilot (Rule 1 example)
• §39177855829 — Smoke Copilot SDK Routed (Rule 1 example)
• §39177856908 — Smoke Pi Routed (Rule 1 example)

Generated by 🔎 Detection Analysis Report · claude · agent · 132.3 AIC · ⌖ 0.855 AIC · ⊞ 6.6K · ◷

• expires on Oct 22, 2026, 3:37 PM UTC-08:00