[detection-analysis] Detection Analysis Report — 2026-11-01

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC) — 2026-10-31T23:20:00Z to 2026-11-01T23:20:00Z
• Total runs analyzed: 645 (across 202 distinct workflows)
• Detection-enabled runs: 557 (86.4%)
• Regular runs (gh-aw-detection: false): 88 (13.6%)
• Misconfigured workflows found: 0

Note

No misconfigured gh-aw-detection workflows were found in this window. Detection ran cleanly across all 618 runs that included a detection job (613 success, 5 skipped, 0 failures), and no workflow alternated between enabled/disabled modes.

Comparison Chart

Metric	Regular Runs	Detection Runs
Run count	88	557
Success rate	75.0%	70.8%
Failure count	22	161
Avg. token usage	101,638	178,092

Misconfigured Workflows

No misconfigured workflows detected in this window.

All three detection rules were checked against the 645 analyzed runs:

1. gh-aw-detection: false on a high-frequency workflow — the 56 workflows running with detection disabled are all low-frequency "Smoke *"/conformance test workflows (≤3 runs each in this 24h window). None crossed the literal >3-run threshold within the window itself; 7-day run counts weren't directly queryable from this 24h log pull, so this check is a conservative proxy (see Recommendations).
2. Detection job itself failing — 0 of 618 runs with a detection job showed a failing job conclusion (613 success, 5 skipped).
3. Inconsistent flag within the window — 0 workflows had both detection-enabled and detection-disabled runs in this window.

View All Run Metrics

Workflows with 2+ runs in the window (39 of 202 total; the remaining 163 workflows each had exactly 1 run):

Workflow	Total Runs	Detection	Regular	Success Rate	Avg Tokens
PR Sous Chef	123	123	0	0.0%	5,360
Issue Monster	47	47	0	97.8%	73,501
Test Quality Sentinel	40	40	0	95.0%	12,070
Design Decision Gate 🏗️	40	40	0	100.0%	125,562
Ponytail Reviewer	39	39	0	100.0%	158,200
Matt Pocock Skills Reviewer	38	38	0	94.7%	47,530
PR Code Quality Reviewer	36	36	0	100.0%	13,259
Daily Go Test Parallelizer	12	12	0	100.0%	1,411,248
AI Moderator	8	8	0	87.5%	919
Auto-Triage Issues	7	7	0	100.0%	136,631
Contribution Check	7	7	0	100.0%	22,005
Smoke Claude Dynamic Workflow	5	5	0	100.0%	1,132
Linter Miner	5	5	0	100.0%	16,984
PR Triage Agent	4	4	0	100.0%	75,172
Avenger	4	4	0	100.0%	1,203,409
[aw] Failure Investigator (6h)	4	4	0	75.0%	332
Smoke Pi Routed	3	0	3	66.7%	47,877
Smoke Copilot MAI	3	0	3	100.0%	75,023
Smoke Copilot Routed	3	3	0	66.7%	62,345
Smoke Codex	3	0	3	100.0%	404,726
Smoke Codex Routed OpenAI	3	0	3	100.0%	30,028
Smoke Claude	3	0	3	66.7%	10,639
Smoke Codex Routed	3	0	3	100.0%	58,362
Smoke OTEL	3	0	3	66.7%	351,206
Smoke Copilot - AOAI (Entra)	3	0	3	0.0%	0
Smoke Claude Routed Anthropic	3	0	3	100.0%	1,900
Smoke Copilot	3	0	3	66.7%	3,811
Smoke Copilot Small	3	0	3	66.7%	2,310
Smoke Gemini	3	3	0	66.7%	3,875,273
Smoke Claude Routed	3	0	3	100.0%	1,834
Smoke Copilot - AOAI (apikey)	3	0	3	33.3%	414,627
Smoke Copilot SDK Routed	3	0	3	100.0%	3,852
Code Scanning Fixer	3	3	0	100.0%	11,459
Smoke Copilot Auto	2	0	2	100.0%	4,462
Smoke Pi	2	0	2	50.0%	41,901
Agent Container Smoke Test	2	0	2	100.0%	204,696
Daily Credit Limit Test	2	2	0	100.0%	47,206
Changeset Generator	2	2	0	100.0%	323,176
Smoke Agy (Experimental)	2	0	2	100.0%	223,534

Coverage note: 191 downloaded run directories lacked aw_info.json (non-executed, cancelled, or metadata-only runs) and were excluded from classification; this does not affect the detection-vs-regular comparison since those runs never reached a classifiable job state.

View Historical Trend

Trend covers 2026-10-03 to 2026-11-01 (22 of those 30 days have data points; 33 distinct days of history exist overall going back to 2026-08-25).

Recommendations

• No detection-feature action required this window. All three misconfiguration rules came back clean — detection ran successfully everywhere it was enabled, and every workflow with detection disabled is a low-frequency smoke/conformance test (≤3 runs/24h), which is a reasonable, likely-intentional use of the opt-out for fast engine-behavior testing rather than a misconfiguration.
• Rule 1 (>3 runs/7 days with detection disabled) could only be approximated from this 24h log pull — a dedicated 7-day query would give a higher-confidence answer for the ~56 workflows currently opting out, in case any of them run more often than their 24h sample suggests.
• Unrelated reliability signal (not a detection issue): PR Sous Chef failed in 122 of 123 runs this window, but job-level attribution shows both its agent and detection jobs succeeded every time — the failures trace to its safe_outputs job (e.g. §41252204558). Worth a separate look outside this detection-focused report.

References:

• §41252204558 — example PR Sous Chef run (safe_outputs job failure, detection/agent jobs succeeded)

Generated by 🔎 Detection Analysis Report · claude · agent · ~103 AIC · ⌖ ~1 AIC · ⊞ 4.4K · ◷

• expires on Nov 4, 2026, 3:32 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 #74254.