[detection-analysis] Detection Analysis Report — 2026-06-28 (last 24h)

[github-actions[bot]]

Summary

• Window evaluated: last available run window (UTC) — 2026-06-28T10:39:48Z → 2026-06-28T12:57:13Z
• Total runs analyzed: 45
• Detection-enabled runs: 42 (93.3%)
• Regular runs: 3 (6.7%)
• Misconfigured workflows found: 0

Note

No workflow misconfigurations were detected against the four criteria. However, one data caveat applies: the documented features.gh-aw-detection flag was absent from every aw_info.json in this window. Detection status was therefore derived from the presence of a detection job in each run's job_details (the actual signal available). See Recommendations.

Comparison Chart

Metric	Regular Runs	Detection Runs
Total runs	3	42
Success rate (%)	100.0	88.1
Avg tokens	0*	0*
Failure count	0	5
Misconfigured count	—	0

* Token usage was not populated in the downloaded run summaries for this window (TokenUsage = 0 across all 45 runs), so token averages are reported as unavailable rather than zero-by-fact.

The 5 detection-run failures (Avenger ×3, Daily Max AI Credits Test (Intentionally Fails) ×1, PR Code Quality Reviewer ×1) all had their detection job conclude success — the run-level failures originate in the agent/safe-output stages, not in detection. One of those (Intentionally Fails) is a deliberate test workflow.

Misconfigured Workflows

No misconfigured workflows detected in this window.

Criterion	Result
1. Detection explicitly false on active workflow (>3 runs/7d)	None — flag never explicitly set to false
2. Audit/report-named workflow missing detection	None — both report workflows (GitHub API Consumption Report Agent, GitHub MCP Remote Server Tools Report Generator) have detection enabled
3. Detection job failures	None — all 42 detection jobs concluded success
4. Inconsistent detection state within 24h	None — every workflow is uniformly detection-on or detection-off

The only workflow lacking a detection job is Smoke CI (3/3 runs, all success, copilot engine). It does not match any misconfiguration criterion: it is a CI smoke test (not an audit/report name), has no explicit false flag, runs ≤3 times, and is consistently detection-off. Flagged here as an observation, not a misconfiguration.

View All Run Metrics

Workflow	Detection	Runs	Success	Fail	Engine
Agent Persona Explorer	✅ all	1	1	0	pi
Auto-Triage Issues	✅ all	1	1	0	pi
Avenger	✅ all	3	0	3	claude
Claude Code User Documentation Review	✅ all	1	1	0	claude
Constraint Solving — Problem of the Day	✅ all	1	1	0	copilot
Daily Documentation Updater	✅ all	1	1	0	pi
Daily Go Function Namer	✅ all	1	1	0	pi
Daily Max AI Credits Test (Intentionally Fails)	✅ all	1	0	1	copilot
Design Decision Gate 🏗️	✅ all	4	4	0	claude
Draft PR Cleanup	✅ all	1	1	0	copilot
GitHub API Consumption Report Agent	✅ all	1	1	0	claude
GitHub MCP Remote Server Tools Report Generator	✅ all	1	1	0	claude
Impeccable Skills Reviewer	✅ all	4	4	0	copilot
Matt Pocock Skills Reviewer	✅ all	4	4	0	copilot
PR Code Quality Reviewer	✅ all	4	3	1	copilot
PR Sous Chef	✅ all	3	3	0	copilot
PR Triage Agent	✅ all	1	1	0	copilot
Package Specification Enforcer	✅ all	1	1	0	pi
Plan Command	✅ all	2	2	0	copilot
Q	✅ all	1	1	0	copilot
Smoke CI	—	3	3	0	copilot
Team Status	✅ all	1	1	0	copilot
Test Quality Sentinel	✅ all	4	4	0	copilot

View Historical Trend

Trend chart skipped: only 2 days of history are stored in cache-memory/trending/detection-metrics/history.jsonl (yesterday + today); the trend chart requires ≥7 days. Today's data point was appended for future runs.

Date	Regular runs	Detection runs	Detection success %
2026-06-27	71	0	—
2026-06-28	3	42	88.1

The jump from 0 → 42 detection runs reflects the detection job now being present in nearly all workflows.

Recommendations

1. Populate features.gh-aw-detection in aw_info.json. The documented metadata field that this analysis is meant to read does not exist in any run. Classification currently relies on indirectly observing a detection job in job_details. Emitting the explicit flag would make detection state unambiguous and machine-checkable.
2. Record token usage in run summaries. TokenUsage was 0 for all 45 runs, blocking the token-comparison metric. Verify the summary pipeline captures token_usage.jsonl into run_summary.json.
3. Smoke CI — confirm whether detection should be intentionally off for the CI smoke test. If detection is desired for parity, add gh-aw-detection: true; otherwise no action needed.
4. Avenger failures — 3/3 Avenger runs failed (detection job succeeded each time). The failures are unrelated to detection but warrant a separate look at its agent/safe-output stages.

References:

• §28320192513 — Avenger (failure)
• §28322748037 — PR Code Quality Reviewer (failure)
• §28322761840 — Claude Code User Documentation Review (detection job success)

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 · 156.9 AIC · ⌖ 35.9 AIC · ⊞ 7.1K · ◷

• expires on Jul 1, 2026, 5:19 AM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Detection Analysis Report.

A newer discussion is available at Discussion #42263.