industrial-ai-troubleshooting-agent: Command for Claude Code

Audit Trail Report

Generate a formatted audit report from a completed root cause analysis trace.

ARGUMENTS: $ARGUMENTS

Conventions

• Python: Always use the project's virtual-environment Python, never bare python (which on macOS may not exist or may point at a dependency-free system Python). Written as <venv-python> below — substitute .venv/bin/python on macOS/Linux or .venv/Scripts/python on Windows.
• Opening files: Use the command for the user's OS, written as <open> below — open on macOS, xdg-open on Linux, start on Windows.

Steps

1. Identify the trace

If the user provided a run_id as the argument, use that.

If no run_id was provided, list available traces:

<venv-python> -c "import os; dirs = sorted(os.listdir('traces'), reverse=True)[:10]; [print(d) for d in dirs]"

If there is only one trace or the user just completed an investigation, use the most recent one. Otherwise ask which trace to use.

2. Ensure corrective actions are logged

Before finalizing the session, confirm the trace contains at least one corrective_action reasoning entry. If the investigation reached a root-cause conclusion but no corrective action was logged, log one or more now with audit_log_reasoning (reasoning_type="corrective_action") before proceeding — each grounded in the confirmed root cause, citing its evidence_steps and the relevant troubleshooting guide / SOP (use docs_search if needed, and cite the document title and revision). Do this now, while the session is still open, so the entries are captured and the "Recommended Corrective Actions" section of summary.md is never empty.

3. Auto-finalize if needed

Check if the session has a session_meta.json. If not, call audit_end_session to finalize it before rendering.

4. Render the reports

<venv-python> render_audit.py traces/{run_id}

Capture the output. This generates both report.md (detailed) and summary.md (executive summary). If the script fails, diagnose and report the error.