[detection-analysis] Detection Analysis Report — Last 24h (2026-10-09)

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC), ending at this workflow's start (~2026-10-09T23:35:00Z)
• Runs with complete logs analyzed: 152 (of 263 run directories found; 111 had only a top-level run.json with no aw_info.json/run_summary.json — see caveat below)
• Detection-enabled runs: 152 (100%)
• Regular runs (gh-aw-detection: false): 0 (0%)
• Misconfigured workflows found: 0

Note

No misconfigured workflows were detected in this window. No workflow in the fully-logged dataset explicitly disables gh-aw-detection, no detection job showed a failing conclusion, and no workflow alternated its detection setting across runs.

Comparison Chart

Because zero runs explicitly opted out of detection in this window, the "Regular Runs" bar is empty (0 runs, 0% success rate by convention) — all activity in this window ran with detection enabled (either explicitly true or, more commonly, left absent/default).

Metric	Regular (opt-out)	Detection (default/enabled)
Run count	0	152
Success rate	0.0%	86.2%
Failure count	0	21
Avg token usage	0	118,401

Misconfigured Workflows

No misconfigured workflows detected in this window.

View All Run Metrics

Top workflows by run count in this window (all classified as Detection, since none set gh-aw-detection: false):

Workflow	Runs	Success Rate	Avg Tokens	Engine(s)
PR Sous Chef	16	100.0%	2,428	pi
PR Code Quality Reviewer	14	100.0%	713,893	copilot
Ponytail Reviewer	14	100.0%	100,278	codex
Test Quality Sentinel	14	100.0%	10,030	copilot
Design Decision Gate 🏗️	13	100.0%	924	pi
Matt Pocock Skills Reviewer	12	91.7%	39,990	copilot
Impeccable Skills Reviewer	12	100.0%	23,173	copilot
Issue Monster	10	100.0%	216,561	pi
Auto-Triage Issues	8	100.0%	106,398	codex
Daily Go Test Parallelizer	4	0.0%	38,606	codex
Avenger	3	0.0%	78,425	codex
Linter Miner	2	50.0%	43,236	copilot (detection absent → default-enabled)
UK AI Operational Resilience	1	100.0%	9,607	copilot (detection absent → default-enabled)
28 other single-run workflows	28	mixed	mixed	mixed

Highest single-run token usage: PR Code Quality Reviewer run §37996967566 at 2,219,240 tokens.

Data gap caveat: 111 of 263 run directories in /tmp/gh-aw/aw-mcp/logs/ contained only a bare run.json (no aw_info.json, no run_summary.json, no job/step data), so detection status and token usage could not be determined for them. Of these, 64 concluded success, 40 failure, and 7 had no conclusion recorded. The largest contributors were Daily Trajectory Grader Implementer (16), [aw] Failure Investigator (6h) (15), PR Sous Chef (14), and AI Moderator (12). This looks like an incomplete log-download for those runs rather than a gh-aw-detection misconfiguration — the agent/detection jobs never materialized in the downloaded data, so Rule 2 (detection job failure) could not be evaluated for them. Flagging as a log-pipeline gap, not a detection misconfiguration.

View Historical Trend

14 days of history now recorded in cache-memory trending store. Detection-run volume has varied widely day to day (152–431 runs), and detection success rate has ranged from ~35% to ~94%, with no sustained trend toward regular (opt-out) usage — it has stayed near 0 on most days.

Recommendations

1. No detection-specific remediation needed this window — the gh-aw-detection feature is effectively universal across active workflows (0 opt-outs observed), so there is no regular-vs-detection drift to correct.
2. Investigate the log-download gap separately from this report's scope: 111/263 runs in the window lack aw_info.json/run_summary.json. If this recurs, it limits future misconfiguration detection (rules 2 and 3 can't be evaluated without job_details). Example incomplete run: §37952640103 (Avenger).
3. Daily Go Test Parallelizer and Avenger show 0% success rate in the fully-logged sample (both detection-enabled, not a detection issue) — worth a separate triage pass on their failure causes, example: §37952461861 (Daily Formal Spec Verifier, also 0% in this window).

References:

• §37952434427 — PR Code Quality Reviewer, detection-enabled example
• §37996967566 — highest token usage run in window
• §37952640103 — example run with incomplete downloaded logs

Generated by 🔎 Detection Analysis Report · claude · agent · 133.1 AIC · ⊞ 6.5K · ◷

• expires on Oct 12, 2026, 3:37 PM UTC-08:00