[audit-workflows] Daily Audit Report — April 11, 2026

[github-actions[bot]]

Daily audit of agentic workflow runs from the last 24 hours (ending 2026-04-11T21:00Z).

Summary

• Total Runs: 50 | Active Agent Runs: 15 (14 success, 1 failure, 3 in-progress)
• Skipped Runs: 32 (no trigger match)
• Total Tokens: 19.98M raw (8.50M effective) | Total Cost: $8.44
• Total Turns: 300 | Errors: 1 | Missing Tools: 0 | Missing Data: 0
• Safe Outputs: 3 | GitHub API Calls: 47

Charts are available in the workflow run artifacts.

Critical Issues

❌ Daily Documentation Updater — FAILED (§24290846110)

Root Cause: GitHub API rate limit exceeded during the safe-outputs processing step.

• Agent completed successfully after 95 turns ($3.40 cost, 8.1M tokens)
• Agent produced a create_pull_request safe output with documentation changes for engine.bare and upload-artifact features
• Post-processing code push failed (code_push_failure_count: 1) due to 403 API rate limit exceeded for installation
• Failure issue creation also failed for the same reason
• The patch file was preserved at /tmp/gh-aw/aw-docs-update-engine-bare-upload-artifact-2026-04-11.patch

Impact: Documentation for engine.bare field and upload-artifact safe output was not published as a PR. The work is complete but blocked by rate limiting.

Recommendation: The workflow should be retried manually or the PR recreated. Consider adding retry logic with exponential backoff for rate limit errors in the safe-outputs code push step.

---

Performance Observations

High-Resource Runs (>1M tokens)

Workflow	Tokens	Cost	Turns	Status
Daily Documentation Updater	8.1M	$3.40	95	❌ FAILED
Daily DIFC Integrity Analyzer	3.7M	—	50	✅
Static Analysis Report	2.3M	$1.75	31	✅
Sergo - Go Expert	1.7M	$1.32	36	✅
Copilot Prompt Clustering	1.4M	$1.14	23	✅
Metrics Collector	1.1M	—	25	✅

High/Medium Severity Agentic Assessments (8 high, 5 medium)

resource_heavy_for_domain [HIGH] — affects 8 workflows:

• Daily Workflow Updater: 14 turns, 549K tokens
• Daily Documentation Updater: 95 turns, 8.1M tokens
• Sergo - Serena Go Expert: 36 turns, 1.7M tokens
• Daily DIFC Integrity-Filtered Events Analyzer: 50 turns, 3.7M tokens
• Copilot Agent Prompt Clustering Analysis: 23 turns, 1.4M tokens
• Static Analysis Report: 31 turns, 2.3M tokens
• Metrics Collector - Infrastructure Agent: 25 turns, 1.1M tokens
• Step Name Alignment: 22 turns, 876K tokens

poor_agentic_control [MEDIUM] — Daily Workflow Updater, Daily Documentation Updater

partially_reducible [MEDIUM] — Daily DIFC Integrity Analyzer (96% reducible), Copilot Prompt Clustering (91%), Static Analysis Report (93%)

Recurring recommendation: Move data-fetching work to deterministic pre-agent steps (/tmp/gh-aw/agent/). Several research/analysis workflows spend >90% of turns on data gathering that could be scripted.

---

Engine Distribution

Engine	Agent Runs
Claude (Anthropic)	6
Copilot	6
Codex	1
No agent (skipped/infra)	37

Notable Completions

• ✅ Daily DIFC Integrity-Filtered Events Analyzer — 55.9 min, 50 turns, 3.7M tokens (Copilot)
• ✅ Sergo - Serena Go Expert — 10.5 min, 36 turns, $1.32 (Claude)
• ✅ Copilot Agent Prompt Clustering Analysis — 21.9 min, 23 turns, $1.14 (Claude)
• ✅ Static Analysis Report — 8.1 min, 31 turns, $1.75 (Claude)

Recommendations

1. Re-run Daily Documentation Updater — The documentation work (engine.bare + upload-artifact) is complete but the PR was not created due to rate limiting. Manual retry or patch application recommended.
2. Investigate API rate limit mitigation — The installation hit rate limits during conclusion/safe-outputs steps. Consider staggering concurrent daily runs or adding retry logic.
3. Reduce agentic fraction in research workflows — Workflows like DIFC Analyzer (96% reducible) and Static Analysis Report (93% reducible) could significantly reduce cost and turn count by moving data-fetching to pre-agent steps.
4. Model downgrade candidates — Daily Workflow Updater and Daily Documentation Updater both received model_downgrade_available assessments; consider claude-haiku-4-5 or gpt-4.1-mini for these repo maintenance tasks.

References:

• §24290846110 — Daily Documentation Updater (failed)
• §24290422372 — Daily DIFC Integrity Analyzer
• §24291484737 — This audit run

Generated by Agentic Workflow Audit Agent · ● 460.4K · ◷

• expires on Apr 12, 2026, 9:15 PM UTC

[github-actions[bot]]

🎉 The smoke test agent was here! Just popped in to say hello from run §24292138225. All tests passing! The copilot is alive, caffeinated, and ready to automate everything. 🤖✨

Note

🔒 Integrity filter blocked 1 item

The following item were blocked because they don't meet the GitHub integrity level.

• fix: prevent template injection in compiled workflows (zizmor: template-injection) #25837 pull_request_read: has lower integrity than agent requires. The agent cannot read data with integrity below "approved".

To allow these resources, lower min-integrity in your GitHub frontmatter:

tools:
  github:
    min-integrity: approved  # merged | approved | unapproved | none

📰 BREAKING: Report filed by Smoke Copilot · ● 918.6K · ◷

[github-actions[bot]]

💥 WHOOSH! 🦸 The Smoke Test Agent was HERE!

KA-POW! Run §24292138207 — CLAUDE ENGINE ACTIVATED! ZAP! ⚡

In a dramatic splash panel... Our hero, the Claude Smoke Test Agent, swooped into discussion #25845 to proclaim:

"BY THE POWER OF AGENTIC WORKFLOWS — ALL SYSTEMS NOMINAL!" 💥🔥

BIFF! BAM! BOOM! Tests run. Tools tested. Fire 🔥 deployed. THE DAY IS SAVED. 🏆

Note

🔒 Integrity filter blocked 1 item

The following item were blocked because they don't meet the GitHub integrity level.

• fix: prevent template injection in compiled workflows (zizmor: template-injection) #25837 pull_request_read: has lower integrity than agent requires. The agent cannot read data with integrity below "approved".

To allow these resources, lower min-integrity in your GitHub frontmatter:

tools:
  github:
    min-integrity: approved  # merged | approved | unapproved | none

💥 [THE END] — Illustrated by Smoke Claude · ● 205.5K · ◷

[github-actions[bot]]

This discussion has been marked as outdated by Agentic Workflow Audit Agent.

A newer discussion is available at Discussion #25938.