[detection-analysis] Detection Analysis Report — 2026-08-02

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC) requested — the log tool caps results at 100 runs, so actual coverage is 2026-08-02T17:33:41Z → 2026-08-02T23:15:08Z (~5.7h) given this repo's high run volume
• Total runs analyzed: 100
• Detection-enabled runs: 89 (89%)
• Regular runs: 11 (11%)
• Misconfigured workflows found: 1

Warning

1 misconfigured workflow detected: a monitoring-role workflow is missing gh-aw-detection: true.

Comparison Chart

Detection-enabled runs had a higher success rate (88.76% vs 81.82%) but over 2x the average token usage (55,068 vs 26,522) — expected overhead from the added threat-detection job.

Misconfigured Workflows

Workflow	Misconfiguration Type	Run Count	Recommended Fix
Daily Regulatory Report Generator	Name suggests audit/monitoring role but gh-aw-detection is not enabled	1	Add gh-aw-detection: true to .github/workflows/daily-regulatory.md frontmatter — this workflow's own description says it "monitors and cross-checks other daily report agents' outputs for data consistency and anomalies," matching the audit/monitor pattern.

Note: Daily Agentic Workflow AIC Usage Audit (.github/workflows/agentic-token-audit.md) was excluded from this check per the documented repository-level opt-out (source-managed from githubnext/agentic-ops).

View All Run Metrics

Workflow	Runs	Detection	Success Rate	Avg Tokens	Engines
PR Sous Chef	11	✅	90.91%	42,034	pi
PR Code Quality Reviewer	10	✅	50.0%	36,271	copilot
Impeccable Skills Reviewer	10	✅	100.0%	7,549	copilot
Design Decision Gate 🏗️	10	✅	70.0%	15,559	claude
Matt Pocock Skills Reviewer	10	✅	100.0%	13,360	copilot
Test Quality Sentinel	10	✅	100.0%	13,587	copilot
Avenger	6	✅	100.0%	9,019	claude
Issue Monster	6	✅	100.0%	29,698	pi
Daily AWF Spec Compiler Surfacing Review	1	✅	100.0%	68,311	pi
ESLint Monster	1	—	100.0%	11,350	pi
Duplicate Code Detector	1	✅	100.0%	2,181,533	codex
Daily Regulatory Report Generator	1	—	100.0%	20,594	copilot
Agentic Workflow Audit Agent	1	✅	100.0%	74,390	claude
Daily Safe Output Tool Optimizer	1	—	100.0%	82,338	claude
Daily Caveman Optimizer	1	✅	100.0%	15,002	claude
Contribution Check	1	✅	100.0%	26,947	copilot
Daily Project Performance Summary Generator (Using MCP Scripts)	1	✅	100.0%	20,537	copilot
Copilot PR Prompt Pattern Analysis	1	✅	100.0%	16,052	copilot
Lockfile Statistics Analysis Agent	1	—	100.0%	15,361	claude
Daily Team Evolution Insights	1	—	100.0%	24,242	claude
Daily Ambient Context Optimizer	1	✅	100.0%	27,488	copilot
Daily Graft Intelligence	1	✅	100.0%	6,645	copilot
Smoke Copilot Sub Agents	1	—	100.0%	18,487	copilot
[aw] Failure Investigator (6h)	1	✅	100.0%	45,070	claude
Daily Safe Output Integrator	1	—	100.0%	8,893	copilot
Discussion Task Miner - Code Quality Improvement Agent	1	—	0.0%	12,693	copilot
Code Scanning Fixer	1	✅	100.0%	8,296	copilot
Daily Code Metrics and Trend Tracking Agent	1	✅	100.0%	47,883	claude
Daily Cache Strategy Analyzer	1	✅	100.0%	770,363	codex
Auto-Triage Issues	1	✅	100.0%	8,627	pi
Copilot Agent PR Analysis	1	✅	100.0%	19,647	claude
PR Triage Agent	1	—	100.0%	7,749	copilot
Daily Testify Uber Super Expert	1	✅	0.0%	6,253	copilot
Linter Miner	1	—	0.0%	78,372	copilot
Daily Secrets Analysis Agent	1	—	100.0%	11,660	copilot

View Historical Trend

30-day history now has 30 recorded entries (23 within the trailing 30-day window). Detection-enabled adoption has grown sharply, overtaking regular runs on 2026-07-30 and again in today's window (89 vs 11).

Recommendations

1. Add gh-aw-detection: true to .github/workflows/daily-regulatory.md — its own description matches the monitoring/audit pattern this repo requires detection for.
2. No detection-step failures were observed among the 89 detection-enabled runs this window — the feature itself is healthy.
3. Detection-enabled runs cost ~2.1x the average tokens of regular runs; if this pattern holds as adoption approaches 100%, consider tracking detection-attributable token overhead separately from agent-task tokens.
4. The 7-day misconfiguration check (rule 1) could only use the same ~5.7h capped window as the 24h check, since the logs tool caps result batches at 100 runs regardless of date range — true 7-day-scoped queries would need date-bounded pagination in the underlying tool to be reliable.

Generated by 🔎 Detection Analysis Report · agent · 142.3 AIC · ⌖ 40.3 AIC · ⊞ 7.2K · ◷

• expires on Aug 5, 2026, 3:39 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 #50116.