[cli-tools-test] bug: audit key_finding incorrectly reports "0 error(s)" for failed workflows with errors

[github-actions]

Problem Description

When a workflow fails and has errors in the errors array, the key_findings description for the "Workflow Failed" finding always shows "failed with 0 error(s)" regardless of how many errors are actually present.

Tool

• Tool: audit

Steps to Reproduce

1. Run audit on a failed workflow run (e.g., run 25954107671 — "Static Analysis Report")
2. Inspect the key_findings array for the entry with title: "Workflow Failed"
3. Observe the description field

Expected Behavior

{
  "title": "Workflow Failed",
  "description": "Workflow 'Static Analysis Report' failed with 2 error(s): ..."
}

Actual Behavior

{
  "title": "Workflow Failed",
  "description": "Workflow 'Static Analysis Report' failed with 0 error(s): ##[error]Process completed with exit code 1."
}

Meanwhile, the errors array at the top level contains 2 entries:

[
  { "file": "agent", "type": "step_failure", "message": "##[error]Process completed with exit code 1." },
  { "file": "agent/Install gh-aw extension", "type": "step_failure", "message": "##[error]Process completed with exit code 1." }
]

Root Cause (hypothesis)

The error count in the key_findings description appears to be computed from a different (possibly empty) source than the errors array that is actually populated in the audit output.

Environment

• Repository: github/gh-aw
• Run ID: 25954578571
• Date: 2026-05-16
• Reproduced on run: 25954107671 (Static Analysis Report, conclusion: failure)

Impact

• Severity: Medium
• Frequency: Always (for failed runs with errors)
• Workaround: Check the errors array directly rather than relying on the key_findings description

References: §25954107671

Generated by 🧪 Daily CLI Tools Exploratory Tester · ● 18.4M · ◷

• expires on May 23, 2026, 6:17 AM UTC