claude-md-improver

To audit and improve CLAUDE.md files, this skill scans your repository directories to evaluate documentation quality against a rigorous rubric, generating targeted diffs to update missing commands, architecture maps, and workflow details for accurate project context.

Claude may hallucinate or perform poorly due to stale or incomplete CLAUDE.md files. Auditing and optimizing your project context documentation ensures Claude has an accurate, up-to-date mental model of your workflows and constraints, reducing errors.

You should update project context documentation when onboarding a new project or after a major refactor. Running an audit ensures your CLAUDE.md files accurately reflect current architecture, commands, and project-specific gotchas for AI agents.

Yes, this skill operates across root, local, and package-specific directories to evaluate command, architecture, and workflow documentation quality. It requires Read, Glob, Grep, Bash, and Edit tool access to perform file discovery and targeted content updates.

The best way to fix poorly structured CLAUDE.md documentation is through automated auditing that evaluates files against a quality rubric, generating actionable diffs to resolve missing commands and architecture maps for optimized AI agent context.

Automated project context auditing requires explicit Read, Glob, Grep, Bash, and Edit tool access to function. It focuses on evaluating and updating CLAUDE.md files and cannot infer unstated architectural constraints without existing text documentation.