[detection-analysis] Detection Analysis Report — 2026-08-28

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC), 2026-08-27T01:59Z → 2026-08-28T01:59Z
• Total runs analyzed: 288
• Detection-enabled runs: 286 (99.3%)
• Regular runs (gh-aw-detection: false): 2 (0.7%)
• Misconfigured workflows found: 0

Note

No misconfigured workflows were detected in this window. The gh-aw-detection feature is behaving consistently across the fleet: all runs with detection enabled that reached the detection job completed it successfully (280 success, 1 skipped, 7 runs where the agent job itself never started and so the job wasn't scheduled).

Comparison Chart

Regular runs show a higher success rate (100%) than detection-enabled runs (75.9%), but this is not attributable to the detection feature itself — see analysis below.

Misconfigured Workflows

No misconfigured workflows detected in this window.

Two workflows explicitly disabled detection (gh-aw-detection: false) but each had only a single run in the 24h window, well under the >3-runs threshold used to flag an active misconfiguration:

Workflow	Run	Status	Tokens
Daily GitHub Docs SEO Optimizer	§33048286684	success	43,123
Smoke Codex	§33116420458	success	45,741

Neither showed alternating detection flags, and no detection job in the window had a failing conclusion (280 success / 1 skipped / 7 not scheduled because the agent job didn't run) — so no rule-2 or rule-3 flags either.

Note: the ">3 runs in the last 7 days" threshold in rule 1 was evaluated against the 24h window only, since only 24h of logs are available to this analysis. Both flagged-candidate workflows should be rechecked against a full 7-day window if detection needs to be audited more strictly.

View All Run Metrics

Top workflows by run count in the window (all runs shown had detection enabled unless noted):

Workflow	Runs	Detection-enabled	Regular	Success %	Avg Tokens
Test Quality Sentinel	20	20	0	100.0%	8,928
Matt Pocock Skills Reviewer	20	20	0	100.0%	13,457
Ponytail Reviewer	19	19	0	100.0%	254,123
Design Decision Gate 🏗️	19	19	0	84.2%	12,287
PR Code Quality Reviewer	18	18	0	100.0%	50,494
Impeccable Skills Reviewer	18	18	0	100.0%	7,814
PR Description Updater	12	12	0	75.0%	9,071
Auto-Triage Issues	6	6	0	50.0%	9,926
Daily Go Test Parallelizer	5	5	0	60.0%	129,533
PR Sous Chef	5	5	0	80.0%	45,013
Contribution Check	5	5	0	60.0%	13,222
Issue Monster	5	5	0	80.0%	63,815
PR Triage Agent	3	3	0	66.7%	36,240
Deep Report	3	3	0	33.3%	29,853
[aw] Failure Investigator (6h)	3	3	0	66.7%	6,357
Code Scanning Fixer	3	3	0	0.0%	11,063
Avenger	3	3	0	0.0%	0
AI Moderator	2	2	0	100.0%	123,278
Release	2	2	0	100.0%	4,821
Daily GitHub Docs SEO Optimizer	1	0	1	100.0%	43,123
Smoke Codex	1	0	1	100.0%	45,741

Remaining 116 workflows each had exactly 1 run in the window (all detection-enabled), omitted here for brevity — full detail is in the run logs.

Aggregate metrics:

Metric	Regular	Detection-enabled
Run count	2	286
Success rate	100.0%	75.9%
Failure count	0	69
Avg tokens/run	44,432	64,205

The lower detection-group success rate reflects that it contains nearly the entire fleet (286 of 288 runs), including many single-run experimental/daily workflows with unrelated agent-level failures — it is not evidence that the detection feature causes failures. No detection job itself failed in this window.

View Historical Trend

Trending history now has 2 daily entries (started 2026-08-25). At least 7 entries are needed before a 30-day trend chart can be generated — check back once more daily runs accumulate.

Recommendations

• No immediate action required — no misconfigurations found this window.
• Keep an eye on Daily GitHub Docs SEO Optimizer and Smoke Codex: both explicitly set gh-aw-detection: false. If their run frequency increases past 3 runs/week, revisit whether the opt-out is still justified.
• Continue accumulating daily trending data; a 30-day trend chart will become available once the history reaches 7 entries.

References:

• §33048286684 — Daily GitHub Docs SEO Optimizer (regular)
• §33116420458 — Smoke Codex (regular)
• §33046665669 — example detection-enabled run

Generated by 🔎 Detection Analysis Report · claude · agent · 107.3 AIC · ⌖ 7.71 AIC · ⊞ 9K · ◷

• expires on Aug 30, 2026, 6:04 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 #56740.