[detection-analysis] Detection Analysis Report — 2026-08-04

[github-actions[bot]]

Summary

• Window evaluated: last 24 full hours (UTC), 2026-08-03T23:38Z to 2026-08-04T23:38Z
• Total runs analyzed: 101 (67 additional runs in the window had no captured agent metadata — logs directory was empty, likely pre-execution failures — and were excluded from classification)
• Detection-enabled runs: 83 (82.2%)
• Regular runs: 18 (17.8%)
• Misconfigured workflows found: 6

Warning

6 misconfigured workflows were detected in this window — see table below.

Comparison Chart

Detection-enabled runs had a higher success rate (95.2%) than regular runs (72.2%), and higher average token usage (36,122 vs 18,946) — consistent with detection runs skewing toward more complex/heavier workflows in this repo.

Misconfigured Workflows

Workflow	Misconfiguration Type	Run Count	Recommended Fix
Smoke Aider	detection-disabled-high-frequency	5	Add gh-aw-detection: true to workflow frontmatter — high-frequency workflow (>3 runs observed) should have detection enabled.
Smoke Goose	detection-disabled-high-frequency	4	Add gh-aw-detection: true to workflow frontmatter — high-frequency workflow (>3 runs observed) should have detection enabled.
Daily Regulatory Report Generator	name-suggests-detection-missing	1	Workflow name implies audit/monitoring intent — add gh-aw-detection: true to frontmatter.
PR Code Quality Reviewer	detection-enabled-run-failed	2	Investigate failed run(s) with detection enabled — check workflow logs for detection-step errors (example run: 30943711376).
Issue Monster	detection-enabled-run-failed	1	Investigate failed run with detection enabled — check workflow logs for detection-step errors (example run: 30946538261).
Daily BYOK Ollama Test	detection-enabled-run-failed	1	Investigate failed run with detection enabled — check workflow logs for detection-step errors (example run: 30957197799).

Note: "Daily Agentic Workflow AIC Usage Audit" was checked against the name-pattern rule but excluded per its documented repository-level opt-out (source-managed from githubnext/agentic-ops).

Note on detection-enabled-run-failed: no dedicated detection-step status is exposed in aw_info.json, so this rule uses overall run conclusion as a proxy for failed detection-enabled runs. Confirm these are genuine detection-step issues (vs. unrelated task failures) before filing fixes.

Note on detection-disabled-high-frequency: run counts here reflect only the 24h window (a lower bound on the 7-day count required by the rule); since 24h ⊆ 7d, any workflow with >3 runs in 24h necessarily has >3 runs in the trailing 7 days, so these are confidently flagged. No workflow alternated between detection-enabled/disabled within the window.

View All Run Metrics

Workflow	Run Count	Detection Enabled	Success Rate	Avg Tokens	Engine(s)
Test Quality Sentinel	10	Yes	100.0%	13,395	copilot
Impeccable Skills Reviewer	9	Yes	100.0%	6,281	copilot
PR Code Quality Reviewer	9	Yes	77.8%	29,326	copilot
Matt Pocock Skills Reviewer	9	Yes	100.0%	12,146	copilot
PR Sous Chef	8	Yes	100.0%	45,548	pi
Design Decision Gate 🏗️	8	Yes	100.0%	9,993	claude
Avenger	5	Yes	100.0%	6,595	claude
Smoke Aider	5	No	60.0%	0	aider
Issue Monster	4	Yes	75.0%	24,130	pi
Smoke Goose	4	No	25.0%	1,738	goose
AI Moderator	2	Yes	100.0%	109,043	codex
Daily Testify Uber Super Expert	1	Yes	100.0%	9,951	copilot
PR Triage Agent	1	No	100.0%	23,183	copilot
Copilot Agent PR Analysis	1	Yes	100.0%	24,924	claude
Auto-Triage Issues	1	Yes	100.0%	12,077	pi
Daily Cache Strategy Analyzer	1	Yes	100.0%	811,841	codex
Daily Code Metrics and Trend Tracking Agent	1	Yes	100.0%	46,421	claude
Daily Safe Output Integrator	1	No	100.0%	11,450	copilot
Discussion Task Miner - Code Quality Improvement Agent	1	No	100.0%	20,197	copilot
[aw] Failure Investigator (6h)	1	Yes	100.0%	21,846	claude
Code Scanning Fixer	1	Yes	100.0%	8,991	copilot
PR Description Updater	1	Yes	100.0%	5,167	copilot
Smoke Copilot Sub Agents	1	No	100.0%	18,874	copilot
Daily Graft Intelligence	1	Yes	100.0%	8,890	copilot
Daily Ambient Context Optimizer	1	Yes	100.0%	33,023	copilot
Daily Team Evolution Insights	1	No	100.0%	23,805	claude
Lockfile Statistics Analysis Agent	1	No	100.0%	23,702	claude
Copilot PR Prompt Pattern Analysis	1	Yes	100.0%	16,503	copilot
Daily Project Performance Summary Generator (Using MCP Scripts)	1	Yes	100.0%	22,516	copilot
Daily Caveman Optimizer	1	Yes	100.0%	15,241	claude
Contribution Check	1	Yes	100.0%	39,157	copilot
Daily Safe Output Tool Optimizer	1	No	100.0%	177,018	claude
Agentic Workflow Audit Agent	1	Yes	100.0%	75,688	claude
Daily Regulatory Report Generator	1	No	100.0%	22,873	copilot
Duplicate Code Detector	1	Yes	100.0%	379,988	codex
ESLint Monster	1	No	100.0%	12,966	pi
Daily AWF Spec Compiler Surfacing Review	1	Yes	100.0%	76,182	pi
Daily BYOK Ollama Test	1	Yes	0.0%	0	copilot
Daily Reliability Review	1	Yes	100.0%	34,071	claude

View Historical Trend

Regular-run volume spiked sharply around 2026-07-21–07-24 (up to ~197/day) then dropped off after 2026-07-29, while detection-run volume has been climbing since late July, peaking at 185 on 2026-08-03. Detection success rate has stayed consistently in the high-80s to high-90s range across the 30-day window.

Recommendations

1. Add gh-aw-detection: true to Smoke Aider and Smoke Goose — both run frequently (5 and 4 runs respectively in this 24h window alone) without detection enabled.
2. Add gh-aw-detection: true to Daily Regulatory Report Generator — its name signals an audit/reporting workflow that should be covered by detection.
3. Investigate the failed detection-enabled runs on PR Code Quality Reviewer, Issue Monster, and Daily BYOK Ollama Test to confirm whether failures are detection-step related or unrelated task errors.
4. No workflows showed inconsistent (alternating) gh-aw-detection settings within the 24h window — continue monitoring for config drift in future runs.

Generated by 🔎 Detection Analysis Report · agent · 131 AIC · ⌖ 39.6 AIC · ⊞ 7.4K · ◷

• expires on Aug 7, 2026, 3:41 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 #50949.