context-auditor

To update CLAUDE.md with new chat conventions, trigger a context audit at the end of your session. The auditor scans the chat for decisions and patterns, then generates structured markdown updates for CLAUDE.md and related SKILL.md files.

A chat context audit identifies gaps between chat decisions and formalized project rules. You should run it at the end of a working session, after fixing a series of bugs, or after discussing a new pattern or convention to sync rules.

To check if project rules cover chat decisions, run an audit to scan CLAUDE.md, agents, and documentation. It detects mismatches and missing conventions, then proposes precise rule updates with proper YAML frontmatter and markdown bodies.

Yes, you can generate SKILL.md entries from chat history by running a context audit. It analyzes the current chat context for accepted decisions and recurring questions, then outputs structured updates to SKILL.md files.

The best way to sync guidelines after introducing a new pattern is to run a context audit. It compares the new chat context against existing project rules and automatically generates updates for CLAUDE.md and agents to close any gaps.

Project rules may not reflect chat bug fixes because the session context was not audited. Running a context audit detects these missing fixes, scans existing rules for omissions, and generates structured updates to project guidelines.