[detection-analysis] Detection Analysis Report — 2026-10-08

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC), approximately 2026-10-07T23:32Z → 2026-10-08T23:32Z
• Total workflow run directories found: 294
• Runs classifiable for detection analysis (have processed aw_info.json): 177
• Runs excluded (no processed agentic-workflow metadata available — status/conclusion only, cannot determine detection flag): 117
• Detection-enabled runs: 176 (99.4% of classifiable)
• Regular runs (gh-aw-detection: false): 1 (0.6% of classifiable)
• Misconfigured workflows found: 0

Note

No misconfigured gh-aw-detection workflows were identified in this window. All three misconfiguration rules (persistent opt-out on a high-volume workflow, a failing detection job, and flag-alternation within the window) came back clean.

Comparison Chart

Detection-enabled runs: 176 runs, 82.4% success rate, 144,909 avg tokens.
Regular runs: 1 run, 0% success rate, 382,645 avg tokens — a single data point (Smoke Agy (Experimental)), not statistically comparable to the 176-run detection cohort.

Misconfigured Workflows

No misconfigured workflows detected in this window.

View All Run Metrics

Workflow	Detection Status	Runs	Success Rate	Avg Tokens
Test Quality Sentinel	Enabled	21	100.0%	8,764
Matt Pocock Skills Reviewer	Enabled	16	75.0%	43,068
Ponytail Reviewer	Enabled	16	100.0%	130,586
Impeccable Skills Reviewer	Enabled	16	93.8%	29,020
PR Sous Chef	Enabled	16	100.0%	2,046
PR Code Quality Reviewer	Enabled	15	86.7%	854,424
Design Decision Gate 🏗️	Enabled	15	100.0%	2,409
Issue Monster	Enabled	12	91.7%	138,366
Auto-Triage Issues	Enabled	8	100.0%	105,768
Visual Regression Checker	Enabled	4	100.0%	7,090
Daily Go Test Parallelizer	Enabled	4	0.0%	35,243
Front Page Copy Guard	Enabled	3	0.0%	15,664
Contribution Check	Enabled	2	100.0%	19,050
Smoke Claude Dynamic Workflow	Enabled	2	0.0%	1,804
Smoke Agy (Experimental)	Disabled	1	0.0%	382,645
Release	Enabled	1	100.0%	10,683
Daily AWF Spec Compiler Surfacing Review	Enabled	1	0.0%	4,912,955
Daily Reliability Review	Enabled	1	0.0%	616,877
...25 other single-run workflows	Enabled	1 each	mixed	mixed

Full dataset: 41 distinct workflows, 177 classifiable runs. Single-run workflows aren't meaningfully comparable on success rate; they're included for completeness.

Excluded from classification (117 runs across ai-moderator, aw-failure-investigator, daily-trajectory-grader-implementer, pr-sous-chef, test-quality-sentinel, smoke-copilot variants, avenger, visual-regression-checker, and others): these run directories only contained the raw GitHub Actions run.json with no aw_info.json/run_summary.json, so the detection flag and token usage could not be recovered. 57 succeeded, 53 failed, 7 were still in-progress/queued at snapshot time.

View Historical Trend

Detection run volume has declined from a peak of ~430 runs/day in mid-September to ~176 today, while detection success rate has climbed from ~35–50% to 82% over the same period. Regular (opted-out) run volume stays consistently low to zero.

Recommendations

1. No action required on gh-aw-detection configuration — the feature is enabled by default across essentially the whole fleet (176/177 classifiable runs) and no workflow shows a failing detection job or inconsistent flag usage in this window.
2. Investigate the 117 runs lacking processed logs separately — roughly even split of success/failure with no recoverable aw_info.json suggests a log-processing or artifact-upload gap (possibly agent jobs failing before the usual artifact upload step) for workflows like ai-moderator, daily-trajectory-grader-implementer, and the smoke-copilot-* family. This is a log-pipeline/observability gap, not a detection misconfiguration, but worth a follow-up if accurate detection-adoption reporting matters.
3. Token-usage outlier: Daily AWF Spec Compiler Surfacing Review consumed ~4.9M tokens in a single failed run — far above the rest of the fleet. Worth a separate token-budget/runaway-loop review independent of the detection analysis.
4. Continue monitoring Smoke Agy (Experimental) — it is the only workflow currently opting out of detection, but at 1 run/7 days it stays well under the >3-runs-in-7-days threshold for flagging, so leave as-is unless its run frequency increases.

References:

• §37848575608 — Smoke Agy (Experimental), the only regular (non-detection) run in window
• §37853193285 — Daily AWF Spec Compiler Surfacing Review, token-usage outlier
• §37825605457 — Release, detection-enabled run with no detection job present (default config, not flagged)

Generated by 🔎 Detection Analysis Report · claude · agent · 184.1 AIC · ⊞ 6.4K · ◷

• expires on Oct 11, 2026, 3:36 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 #67301.