[detection-analysis] Detection Analysis Report — 2026-07-20 (last 24h)

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC) — observed run span 2026-07-20T13:16:30Z → 2026-07-20T23:14:05Z
• Total runs analyzed: 161
• Detection-enabled runs: 54 (33.5%)
• Regular runs: 107 (66.5%)
• Misconfigured workflows found: 5

Warning

5 workflows were flagged as potentially misconfigured for the gh-aw-detection feature. See the table below for details and recommended fixes. Four are name-heuristic flags (name implies an analysis/monitoring role but detection is not enabled); one is a real detection-job failure.

Classification uses features.gh-aw-detection from each run's aw_info.json directly (true → detection-enabled; false/absent/unset → regular).

Comparison Chart

Detection-enabled runs had a lower success rate (90.7% vs 97.2%) and higher average token usage (68,602 vs 28,459) than regular runs over the window — expected, since detection adds a post-agent analysis job and tends to run on heavier analytical workflows.

Metric	Regular Runs	Detection Runs
Run count	107	54
Success rate	97.2%	90.7%
Failures	3	5
Avg tokens	28,459	68,602

Average token figures are influenced by a few outliers (e.g. Daily Cache Strategy Analyzer ≈2.18M tokens with detection on; Smoke Copilot ≈965K without).

Misconfigured Workflows

Workflow	Misconfiguration type	Runs	Recommended fix
Daily Regulatory Report Generator	Name implies detection (report) but gh-aw-detection not enabled	1	Add gh-aw-detection: true to frontmatter, or document an intentional opt-out.
Daily Security Observability Report	Name implies detection (report) but gh-aw-detection not enabled	1	Add gh-aw-detection: true — a security report especially should run threat detection.
Deep Report	Name implies detection (report) but gh-aw-detection not enabled	1	Add gh-aw-detection: true, or rename if it is not an analysis/reporting workflow.
MCP Inspector Agent	Name implies detection (inspector) but gh-aw-detection not enabled	1	Add gh-aw-detection: true; this run also failed (conclusion=failure) — investigate separately.
Daily Code Metrics and Trend Tracking Agent	Detection enabled but the detection job failed	1	Investigate run 29770233192 — threat detection did not complete; the overall run also failed.

No workflows triggered the explicit-false-on-active-workflow rule (only Smoke Copilot and Smoke Copilot Sub Agents set gh-aw-detection: false, each with a single run in-window), and no workflow alternated its detection flag within the window.

View All Run Metrics

Per-workflow breakdown (72 distinct workflows). Detection column: ✅ on = all runs detection-enabled, ⚠️ mixed, — off = none enabled.

Workflow	Runs	Detection	Success %	Avg tokens	Failures
PR Sous Chef	16	— off	100%	46,739	0
Matt Pocock Skills Reviewer	13	— off	100%	13,937	0
Design Decision Gate 🏗️	12	— off	100%	12,237	0
Test Quality Sentinel	12	✅ on	100%	10,913	0
Impeccable Skills Reviewer	12	— off	100%	7,632	0
PR Code Quality Reviewer	12	— off	100%	15,386	0
Tidy	8	— off	100%	5,633	0
Issue Monster	5	— off	100%	17,736	0
Contribution Check	4	✅ on	100%	21,936	0
Discussion Task Miner - Code Quality Improvement Agent	2	— off	100%	13,520	0
[aw] Failure Investigator (6h)	2	✅ on	100%	20,614	0
Code Scanning Fixer	2	✅ on	100%	4,954	0
Avenger	2	✅ on	100%	0	0
Daily File Diet	1	✅ on	100%	15,710	0
Agent Performance Analyzer - Meta-Orchestrator	1	✅ on	100%	19,473	0
Workflow Normalizer	1	— off	100%	12,275	0
Repository Quality Improvement Agent	1	— off	100%	26,491	0
CI Optimization Coach	1	✅ on	0%	0	1
CLI Consistency Checker	1	✅ on	100%	53,391	0
Daily Malicious Code Scan Agent	1	✅ on	100%	11,832	0
Daily Fact	1	✅ on	100%	97,288	0
Dead Code Removal Agent	1	— off	100%	13,109	0
Architecture Guardian	1	✅ on	100%	4,061	0
Daily Agent of the Day Blog Writer	1	✅ on	100%	20,648	0
Daily Issues Report Generator	1	✅ on	0%	0	1
PR Description Updater	1	— off	100%	8,324	0
Deep Report	1	— off	100%	26,602	0
Issue Triage Agent	1	— off	100%	3,695	0
Delight	1	— off	100%	17,253	0
Daily CLI Performance Agent	1	✅ on	100%	17,563	0
Weekly Issue Summary	1	— off	100%	10,530	0
Agentic Workflow AIC Usage Optimizer	1	— off	100%	30,768	0
Breaking Change Checker	1	✅ on	100%	8,047	0
Daily Copilot PR Merged Report	1	✅ on	100%	10,539	0
The Daily Repository Chronicle	1	✅ on	100%	14,904	0
Daily Formal Spec Verifier	1	✅ on	0%	14,136	1
UK AI Operational Resilience	1	— off	100%	17,056	0
Agent Persona Explorer	1	✅ on	100%	53,962	0
Daily SPDD Spec Planner	1	— off	100%	12,795	0
Daily Security Observability Report	1	— off	100%	14,132	0
Repository Tree Map Generator	1	— off	100%	11,117	0
Workflow Portfolio Analyst	1	— off	0%	29,536	1
Slide Deck Maintainer	1	— off	100%	15,005	0
GEO Optimizer Daily Audit	1	✅ on	100%	9,216	0
Smoke Copilot	1	— off	0%	964,556	1
Release	1	— off	100%	8,741	0
Linter Miner	1	— off	100%	18,456	0
Daily Secrets Analysis Agent	1	— off	100%	13,541	0
Copilot Opt	1	✅ on	100%	13,876	0
Daily Testify Uber Super Expert	1	✅ on	100%	7,399	0
PR Triage Agent	1	— off	100%	12,537	0
Copilot Agent PR Analysis	1	✅ on	100%	16,898	0
Auto-Triage Issues	1	✅ on	100%	8,932	0
Daily Cache Strategy Analyzer	1	✅ on	100%	2,181,563	0
Daily Code Metrics and Trend Tracking Agent	1	✅ on	0%	31,622	1
MCP Inspector Agent	1	— off	0%	0	1
Daily Safe Output Integrator	1	— off	100%	8,637	0
AI Moderator	1	✅ on	100%	148,240	0
Smoke Copilot Sub Agents	1	— off	100%	27,121	0
Daily Ambient Context Optimizer	1	✅ on	100%	25,392	0
Daily Team Evolution Insights	1	— off	100%	35,723	0
Lockfile Statistics Analysis Agent	1	— off	100%	13,658	0
Copilot PR Prompt Pattern Analysis	1	✅ on	100%	11,748	0
Daily Project Performance Summary Generator (Using MCP Scripts)	1	✅ on	100%	18,987	0
Daily Caveman Optimizer	1	✅ on	100%	12,332	0
Agentic Workflow Audit Agent	1	✅ on	100%	37,951	0
Daily Regulatory Report Generator	1	— off	100%	16,435	0
Duplicate Code Detector	1	✅ on	100%	113,407	0
ESLint Monster	1	— off	100%	21,610	0
Daily AWF Spec Compiler Surfacing Review	1	✅ on	100%	83,655	0
Daily BYOK Ollama Test	1	✅ on	0%	0	1
Daily Reliability Review	1	✅ on	100%	28,902	0

View Historical Trend

Trend across the 2 recorded snapshots (2026-07-16 → 2026-07-20): both run volumes grew, and regular-run success rate recovered sharply (78.9% → 97.2%) while detection success rate held steady (~90%). Misconfigured count is unchanged at 5.

Recommendations

1. Enable detection on reporting/inspection workflows. The four name-heuristic flags (Daily Regulatory Report Generator, Daily Security Observability Report, Deep Report, MCP Inspector Agent) either produce analytical output or inspect repository state — add gh-aw-detection: true to their frontmatter, or add a documented opt-out if detection is intentionally excluded.
2. Fix the failing detection job on Daily Code Metrics and Trend Tracking Agent (run 29770233192): both the detection job and the overall run failed. Confirm the detection engine/credentials are configured before it silently ships unscanned output.
3. Watch the detection success gap. Detection runs trail regular runs by ~6.5 points on success rate; most detection failures coincide with zero-token runs (e.g. Daily BYOK Ollama Test, Daily Issues Report Generator, CI Optimization Coach), suggesting agent-execution failures upstream of detection rather than detection itself.

References:

• §29770233192 — Daily Code Metrics (detection job failed)
• §29770927744 — MCP Inspector Agent (run failed, detection off)
• §29759100274 — Daily Security Observability Report

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• awmgmcpg

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "awmgmcpg"

See Network Configuration for more information.

Generated by 🔎 Detection Analysis Report · 181 AIC · ⌖ 37.6 AIC · ⊞ 6.9K · ◷

• expires on Jul 23, 2026, 3:46 PM UTC-08:00