[detection-analysis] Detection Analysis Report — 2026-09-13 #60704

2026-09-13T23:37:58Z

github-actions[bot]
Bot Sep 13, 2026

Summary

Window evaluated: last 24 full hours (UTC), 2026-09-12T23:19Z → 2026-09-13T23:19Z
Total runs analyzed: 405 (with complete engine/job telemetry); 167 additional runs had no aw_info.json/run_summary.json (agent job never executed — e.g. activation-only or skipped triggers) and are excluded from the detection comparison below
Detection-enabled runs: 364 (89.9%)
Regular runs (gh-aw-detection: false): 41 (10.1%)
Misconfigured workflows found: 0

Note

No misconfigured gh-aw-detection workflows were found in this window. All 41 "regular" runs are one-off smoke-test workflows (e.g. Smoke Pydantic AI, Smoke Aider, Smoke OpenCode) intentionally exercising engines with detection off; none exceeded the >3-run threshold, none showed a failing detection job while enabled, and no workflow flip-flopped its gh-aw-detection setting within the window.

Comparison Chart

Regular runs skew heavily toward one-off engine smoke tests (low n, low observed success rate), while detection runs represent the bulk of production scheduled/triggered workflows. Token usage could not be compared — every run in this log snapshot reports TokenUsage: 0 in run_summary.json, which looks like a metrics-pipeline gap upstream of this analysis rather than an actual detection-vs-regular difference (worth a separate look, see Recommendations).

Misconfigured Workflows

No misconfigured workflows detected in this window.

View All Run Metrics (405 runs across 173 workflows)

Workflow	Total Runs	Detection Runs	Regular Runs	Success Rate	Engine(s)
PR Sous Chef	80	80	0	31.2%	pi
Issue Monster	46	46	0	21.7%	pi
Daily Go Test Parallelizer	12	12	0	91.7%	codex
Design Decision Gate 🏗️	11	11	0	100.0%	pi
Impeccable Skills Reviewer	11	11	0	100.0%	copilot
Matt Pocock Skills Reviewer	11	11	0	100.0%	copilot
PR Code Quality Reviewer	11	11	0	0.0%	copilot
Ponytail Reviewer	11	11	0	100.0%	codex
Test Quality Sentinel	11	11	0	100.0%	copilot
Avenger	10	10	0	90.0%	codex
Auto-Triage Issues	7	7	0	100.0%	codex
Contribution Check	5	5	0	100.0%	copilot
Code Scanning Fixer	4	4	0	0.0%	copilot
Deep Report	4	4	0	100.0%	claude
PR Triage Agent	4	4	0	0.0%	copilot
[aw] Failure Investigator (6h)	4	4	0	75.0%	claude
Release	3	3	0	100.0%	copilot
Smoke Copilot	3	0	3	0.0%	copilot
AI Moderator	2	2	0	0.0%	pi
Daily Credit Limit Test	2	2	0	50.0%	pi
Agent Container Smoke Test	1	0	1	0.0%	pi
Agent Job Health Monitor	1	1	0	0.0%	claude
Agent Performance Analyzer - Meta-Orchestrator	1	1	0	100.0%	copilot
Agent Persona Explorer	1	1	0	100.0%	codex
Agentic Workflow Audit Agent	1	1	0	100.0%	codex
Artifacts Usage Report	1	1	0	100.0%	codex
CLI Version Checker	1	1	0	100.0%	pi
Claude Code User Documentation Review	1	1	0	100.0%	claude
Code Simplifier	1	1	0	0.0%	copilot
Constraint Solving — Problem of the Day	1	1	0	100.0%	copilot
Copilot Agent PR Analysis	1	1	0	0.0%	claude
Copilot Agent Prompt Clustering Analysis	1	1	0	0.0%	claude
Copilot CLI Deep Research Agent	1	1	0	100.0%	copilot
Copilot PR Prompt Pattern Analysis	1	1	0	100.0%	copilot
Copilot Session Insights	1	1	0	100.0%	claude
Daily A/B Testing Advisor	1	1	0	0.0%	codex
Daily AWF Spec Compiler Surfacing Review	1	1	0	100.0%	codex
Daily Ambient Context Optimizer	1	1	0	100.0%	copilot
Daily Assign Issue To User	1	1	0	100.0%	copilot
Daily AstroStyleLite Markdown Spellcheck	1	1	0	100.0%	claude
Daily CLI Performance Agent	1	1	0	100.0%	codex
Daily Cache Strategy Analyzer	1	1	0	100.0%	codex
Daily Caveman Optimizer	1	1	0	100.0%	claude
Daily Cli Tools Tester	1	1	0	100.0%	codex
Daily Code Debt Cleanup — Aider	1	1	0	0.0%	aider
Daily Code Metrics and Trend Tracking Agent	1	1	0	0.0%	copilot
Daily Community Attribution Updater	1	1	0	0.0%	copilot
Daily Compiler Quality Check	1	1	0	100.0%	copilot
Daily Compiler Threat Spec Optimizer	1	1	0	100.0%	copilot
Daily Container Image Security Scan	1	1	0	0.0%	copilot
Daily Documentation Diagram	1	1	0	100.0%	codex
Daily Documentation Healer	1	1	0	100.0%	claude
Daily Documentation Updater	1	1	0	0.0%	codex
Daily Evals Feature Report	1	1	0	100.0%	codex
Daily Formal Spec Verifier	1	1	0	100.0%	copilot
Daily GitHub Docs SEO Optimizer	1	0	1	100.0%	copilot
Daily Go Function Namer	1	1	0	100.0%	pi
Daily Go Test Stubs — Aider	1	1	0	0.0%	aider
Daily Grader Audit	1	1	0	100.0%	claude
Daily Graft Intelligence	1	1	0	100.0%	copilot
Daily Issues Report Generator	1	1	0	100.0%	copilot
Daily Malicious Code Scan Agent	1	1	0	100.0%	copilot
Daily Max Ai Credits Test	1	1	0	100.0%	codex
Daily Model Inventory Checker	1	1	0	100.0%	copilot
Daily Observability Report for AWF Firewall and MCP Gateway	1	1	0	100.0%	codex
Daily PR Code Quality Review	1	1	0	100.0%	copilot
Daily Project Performance Summary Generator (Using MCP Scripts)	1	1	0	100.0%	pi
Daily Regression Audit	1	1	0	100.0%	copilot
Daily Regulatory Report Generator	1	1	0	100.0%	codex
Daily Rendering Scripts Verifier	1	1	0	0.0%	claude
Daily SPDD Spec Planner	1	1	0	100.0%	copilot
Daily Safe Output Integrator	1	1	0	100.0%	copilot
Daily Safe Outputs Conformance Checker	1	1	0	100.0%	claude
Daily Safe Outputs Git Simulator	1	1	0	100.0%	claude
Daily Schema Consistency Audit	1	1	0	100.0%	copilot
Daily Secrets Analysis Agent	1	1	0	100.0%	copilot
Daily Security Observability Report	1	1	0	0.0%	copilot
Daily Spec Coverage Review	1	1	0	100.0%	copilot
Daily Spending Forecast	1	1	0	100.0%	codex
Daily Storify	1	1	0	100.0%	codex
Daily Sub-Agent Model Resolution Audit	1	1	0	100.0%	copilot
Daily Syntax Error Quality Check	1	1	0	100.0%	copilot
Daily Team Evolution Insights	1	1	0	100.0%	goose
Daily Trajectory Grader Implementer	1	1	0	100.0%	copilot
Daily VulnHunter Scan	1	1	0	100.0%	claude
Daily Windows Defender Release Scan	1	1	0	100.0%	codex
Daily Windows Terminal Integration Builder	1	1	0	100.0%	codex
Daily Workflow Updater	1	1	0	100.0%	copilot
Daily action/setup/* Security Audit	1	1	0	100.0%	copilot
Dead Code Removal Agent	1	1	0	0.0%	copilot
Delight	1	1	0	100.0%	copilot
Detection Analysis Report	1	1	0	0.0%	claude
Dev	1	1	0	0.0%	codex
Dictation Prompt Generator	1	1	0	100.0%	copilot
Documentation Noob Tester	1	1	0	100.0%	copilot
Documentation Unbloat	1	1	0	100.0%	pi
Draft PR Cleanup	1	1	0	100.0%	copilot
Duplicate Code Detector	1	1	0	0.0%	pi
ESLint Miner	1	1	0	0.0%	copilot
ESLint Monster	1	1	0	100.0%	codex
ESLint Refiner	1	1	0	100.0%	claude
GPL Dependency Cleaner (gpclean)	1	1	0	0.0%	codex
GitHub API Consumption Report Agent	1	1	0	0.0%	codex
GitHub MCP Remote Server Tools Report Generator	1	1	0	0.0%	claude
GitHub Remote MCP Authentication Test	1	1	0	100.0%	codex
Go Logger Enhancement	1	1	0	100.0%	claude
Instructions Janitor	1	1	0	100.0%	claude
Issue Arborist	1	1	0	100.0%	codex
LintMonster	1	1	0	100.0%	codex
Linter Miner	1	1	0	100.0%	copilot
Lockfile Statistics Analysis Agent	1	1	0	100.0%	claude
Metrics Collector	1	1	0	100.0%	codex
Multi-Device Docs Tester	1	1	0	100.0%	pi
Outcome Collector	1	1	0	100.0%	codex
Package Specification Enforcer	1	1	0	100.0%	codex
Package Specification Extractor	1	1	0	100.0%	copilot
PureLock	1	1	0	0.0%	codex
Schema Consistency Checker	1	1	0	100.0%	codex
Semantic Function Refactoring	1	1	0	100.0%	claude
Sergo - Serena Go Expert	1	1	0	100.0%	claude
Smoke Agent: all/merged	1	0	1	100.0%	claude
Smoke Agent: all/none	1	0	1	100.0%	claude
Smoke Agent: public/approved	1	0	1	100.0%	claude
Smoke Agent: public/none	1	0	1	100.0%	claude
Smoke Agent: scoped/approved	1	0	1	0.0%	claude
Smoke Aider	1	0	1	0.0%	aider
Smoke Call Workflow	1	0	1	100.0%	codex
Smoke Checkout PR Dispatch	1	0	1	100.0%	copilot
Smoke Claude	1	0	1	0.0%	claude
Smoke Claude on Copilot	1	1	0	0.0%	claude
Smoke Codex	1	0	1	100.0%	codex
Smoke Copilot - AOAI (Entra)	1	0	1	0.0%	copilot
Smoke Copilot - AOAI (apikey)	1	0	1	0.0%	copilot
Smoke Copilot ARM64	1	0	1	0.0%	copilot
Smoke Copilot Auto	1	0	1	100.0%	copilot
Smoke Copilot MAI	1	0	1	0.0%	copilot
Smoke Copilot SDK	1	0	1	100.0%	copilot
Smoke Copilot Small	1	0	1	0.0%	copilot
Smoke Copilot Sub Agents	1	0	1	100.0%	copilot
Smoke Create Cross-Repo PR	1	0	1	0.0%	copilot
Smoke Crush	1	0	1	0.0%	crush
Smoke Cursor	1	0	1	0.0%	cursor
Smoke DeepSeek Harness	1	0	1	0.0%	deepseek-harness
Smoke Drive	1	0	1	0.0%	codex
Smoke Gemini	1	1	0	0.0%	gemini
Smoke GitHub Claude	1	1	0	0.0%	claude
Smoke Goose	1	0	1	0.0%	goose
Smoke Issues	1	1	0	0.0%	copilot
Smoke Kiro	1	0	1	0.0%	kiro
Smoke Multi PR	1	0	1	100.0%	copilot
Smoke OTEL	1	0	1	100.0%	codex
Smoke OpenCode	1	0	1	0.0%	opencode
Smoke Pi	1	0	1	100.0%	pi
Smoke Project	1	0	1	0.0%	copilot
Smoke Pydantic AI	1	0	1	0.0%	pydantic-ai
Smoke Service Ports	1	0	1	0.0%	pi
Smoke Temporary ID	1	0	1	0.0%	pi
Smoke Update Cross-Repo PR	1	0	1	0.0%	copilot
Smoke Workflow Call	1	0	1	100.0%	copilot
Smoke Workflow Call with Inputs	1	0	1	100.0%	copilot
Squad Game Planner	1	1	0	100.0%	copilot
Static Analysis Report	1	1	0	100.0%	claude
Step Name Alignment	1	1	0	100.0%	claude
Sub-Issue Closer	1	1	0	0.0%	pi
Team Status	1	1	0	100.0%	copilot
Terminal Stylist	1	1	0	0.0%	pi
The Great Escapi	1	1	0	100.0%	copilot
Tidy	1	1	0	100.0%	copilot
Workflow Health Manager - Meta-Orchestrator	1	1	0	100.0%	copilot
Workflow Normalizer	1	1	0	0.0%	codex
arXiv Paper Researcher: GitHub Agentic Workflows	1	1	0	100.0%	copilot
daily-experiment-report	1	1	0	100.0%	copilot
jsweep - JavaScript Unbloater	1	1	0	100.0%	copilot

View Historical Trend

Detection-run volume has climbed steadily from ~250-430/day over the past ~3 weeks while regular (opted-out) runs stay near-zero most days, spiking only on days with heavier smoke-test activity (e.g. 2026-09-11: 12 regular runs). Detection success rate has been trending down recently (100% on 2026-09-10 → 35.1% on 2026-09-12 → 60.2% today), which tracks a broader rise in agent job failures across the fleet, not a detection-specific regression.

Recommendations

Investigate the zero-token metrics gap. Every run in this snapshot reports TokenUsage: 0 regardless of detection status, which breaks any token-cost comparison between regular and detection runs. Check whether the log-download/processing pipeline that produces run_summary.json is dropping TokenUsage before these reports are generated.
Watch the declining detection success rate. Detection-run success dropped to 35.1% on 2026-09-12 before recovering to 60.2% today — worth confirming this tracks a known incident (e.g. PR Sous Chef at 31.2% and Issue Monster at 21.7% success over this window are large contributors) rather than a detection-job regression.
No action needed on gh-aw-detection configuration — current opt-outs are all intentional, low-volume smoke tests. Re-run this check after 7 days of continuous logs to properly evaluate Rule 1 (>3 runs/7 days), since this window only covers 24h.

Warning

Firewall blocked 1 domain

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

api.anthropic.com

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

network:
  allowed:
    - defaults
    - "api.anthropic.com"

See Network Configuration for more information.

Generated by 🔎 Detection Analysis Report · claude · agent · 162.7 AIC · ⌖ 9.85 AIC · ⊞ 9K · ◷

expires on Sep 16, 2026, 3:37 PM UTC-08:00

Navigation Menu

Uh oh!

[detection-analysis] Detection Analysis Report — 2026-09-13 #60704

Uh oh!

{{title}}

Uh oh!

Replies: 0 comments

Select a reply

Uh oh!

Uh oh!

[detection-analysis] Detection Analysis Report — 2026-09-13 #60704

Uh oh!

github-actions[bot] Bot Sep 13, 2026

Summary

Comparison Chart

Misconfigured Workflows

Recommendations

Replies: 0 comments

github-actions[bot]
Bot Sep 13, 2026