Dev10x:skill-audit

Auditing Claude Code session transcripts involves analyzing visible context or entire transcripts to check for skill compliance, missed invocations, and user corrections. This automated process requires parallel subagent fan-out for detailed analysis and user interaction for disposition.

Automating process improvement for AI workflows involves applying session audits to check for missed invocations and user corrections. This provides insights for continuous refinement of AI workflows by automatically reviewing the workflow to identify areas for enhancement or correction.

Checking for missed invocations and user corrections in a Claude Code session requires analyzing the session transcripts. An automated audit can review the visible context or full transcript to identify workflow deviations and areas for enhancement.

Lightweight session analysis works for routine compliance review by offering a default analysis of visible context. For a detailed forensic audit, you can specify a full transcript analysis which utilizes parallel subagent fan-out to thoroughly check for missed invocations and corrections.

You should not use a full forensic audit when a lightweight default analysis of visible context is sufficient. Reserve full transcript analysis with parallel subagent fan-out for complex sessions requiring deep investigation into missed invocations and user corrections.