[detection-analysis] Detection Analysis Report — 2026-11-09 (partial coverage)

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC) — partial coverage only, see caveat below
• Total runs analyzed: 341 (across 80 distinct workflows)
• Detection-enabled runs: 341 (100%)
• Regular runs (gh-aw-detection: false): 0 (0%)
• Misconfigured workflows found: 0

Warning

Log collection for this run only reached 2026-11-09T11:55Z → 2026-11-09T23:13Z (~11.3 hours), not the full requested 24-hour window (~2026-11-08T23:14Z → 2026-11-09T23:xxZ). A logs MCP continuation call to backfill the missing ~12.7 hours was attempted but hit the per-call count limit after collecting only ~2 additional hours (2026-11-09T12:27Z–14:38Z), with further continuation cursors remaining unresolved. This report covers the analyzed subset only — it does not represent the full 24-hour window.

This gap very likely explains why 0 regular runs were observed today: historical trend data (see below) shows regular runs (gh-aw-detection: false) occurring every prior day in this window (1–55/day), typically from smoke-* and engine-conformance-* workflows that trigger on pull_request events — none of which happened to fall inside the ~11.3h slice that was actually captured. Do not read "0 regular runs" or "0 misconfigured" as evidence those workflows are healthy or absent — they were simply outside the captured sample today.

Comparison Chart

Note: the "Regular Runs" bar is empty because no gh-aw-detection: false runs fell inside the captured ~11.3h sample (see coverage caveat above), not because none exist.

Misconfigured Workflows

No misconfigured workflows detected in this window (within the analyzed ~11.3h sample). Specifically:

• Rule 1 (gh-aw-detection: false with >3 runs/window): no runs with an explicit false value appeared in the captured sample at all.
• Rule 2 (detection job itself failing): 0 of 341 runs had a failing detection job conclusion.
• Rule 3 (alternating enabled/disabled): not triggered.

One non-detection issue worth flagging separately (out of scope for gh-aw-detection rules, since it's not a detection-job failure): the Auditor workflow failed in all 11 of its runs this window (example run §42712106891). In each case the agent job itself was skipped (0 tokens used), consistent with the collect_failure_evidence / "Collect health failures, cached logs and bounded audits" step failing before the agent could start (driver-exit failure, part of a cluster spanning 8 runs in the wider failure investigation). This looks like a trigger/permission or upstream-step issue, not a detection misconfiguration — recommend investigating the collect_failure_evidence step directly.

View All Run Metrics

341 runs across 80 workflows were classified; top 25 by run count:

Workflow	Total Runs	Detection	Regular	Success %	Avg Tokens	Engines
PR Sous Chef	59	59	0	98.3	1,191	pi
Ponytail Reviewer	31	31	0	96.8	227,960	codex
Matt Pocock Skills Reviewer	30	30	0	76.7	82,669	copilot
Test Quality Sentinel	30	30	0	93.3	18,471	copilot
Design Decision Gate 🏗️	29	29	0	86.2	14,471	pi
PR Code Quality Reviewer	29	29	0	82.8	18,302	copilot
PR Specification Reviewer	27	27	0	85.2	2,913	copilot
Auditor	11	11	0	0.0	0	copilot
Auto-Triage Issues	6	6	0	100.0	225,250	codex
Daily Go Test Parallelizer	6	6	0	33.3	1,010,283	codex
AI Moderator	4	4	0	100.0	714	pi
Issue Monster	4	4	0	100.0	65,966	pi
Contribution Check	3	3	0	100.0	43,827	copilot
Visual Regression Checker	3	3	0	100.0	8,160	copilot
Agent Job Health Monitor	2	2	0	50.0	59,151	claude
Avenger	2	2	0	100.0	18,130	copilot
PR Triage Agent	2	2	0	50.0	75,300	copilot
Actions and AW Linter Factory	1	1	0	100.0	374,828	copilot
Agent Performance Analyzer - Meta-Orchestrator	1	1	0	100.0	18,352	copilot
Agent Persona Explorer	1	1	0	100.0	60,296	claude
Agentic Workflow AIC Usage Optimizer	1	1	0	100.0	13,180	copilot
Agentic Workflow Audit Agent	1	1	0	0.0	4,041,032	codex
Architecture Guardian	1	1	0	100.0	7,782	copilot
Breaking Change Checker	1	1	0	0.0	0	copilot
CI Optimization Coach	1	1	0	100.0	15,787	copilot

Aggregate: 341 detection-enabled, 0 regular; 83.0% detection success rate (58 failures); average 86,363 tokens/run (detection group).

View Historical Trend

41 historical daily data points (27 distinct days over the trailing window). Detection success rate has averaged ~85%, dipped to ~67% around Oct 30–Nov 1, and recovered to 83–90% over the past week. Detection run volume has risen from ~200/day in mid-October to 340–650/day recently. Regular-run counts and misconfigured-workflow counts have historically been non-zero most days (e.g. 2026-11-08: 1 regular run, 6 misconfigured; 2026-11-06: 5 regular runs, 1 misconfigured) — reinforcing that today's zero counts reflect the partial-coverage gap, not a real change.

Recommendations

1. Re-run log collection for the missing ~12.7h window (2026-11-08T23:14Z–2026-11-09T11:55Z) before trusting a "0 misconfigured" conclusion for today — the historical pattern strongly suggests smoke-*/engine-conformance-* regular runs and at least one misconfiguration exist in that gap.
2. Investigate the Auditor workflow's collect_failure_evidence step — it failed in all 11 runs this window, preventing the agent job from ever starting (§42712106891).
3. No gh-aw-detection-specific action needed based on the captured sample, but re-check once full-window coverage is available.

References:

• §42712106891 — Auditor failure example

Generated by 🔎 Detection Analysis Report · claude · agent · ~104 AIC · ⌖ ~1 AIC · ⊞ 4.5K · ◷

• expires on Nov 12, 2026, 3:30 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 #76630.