claude-md-doctor

To diagnose Claude Code project instructions, audit your CLAUDE.md files and .claude/rules/ directory by evaluating file structure, quality scoring, anti-patterns, staleness, token efficiency, rule modularity, frontmatter correctness, and actionability to generate detailed diagnostic reports with severity ratings and specific fixes.

An AI assistant may ignore project context due to poorly written or outdated project instructions like CLAUDE.md files. Auditing these files for clarity, accuracy, and efficiency ensures the AI has the best possible information for optimal code generation.

You can audit .claude/rules/ by running a diagnostic assessment that checks rule files for modularity, correctness, and focus. This evaluates rule structure and frontmatter correctness to provide specific fixes for any identified anti-patterns.

The best way to improve token efficiency in CLAUDE.md files is to evaluate them for staleness and anti-patterns using a diagnostic tool. This assesses file structure and actionability to provide specific fixes that optimize token usage.

Yes, you can get severity ratings for CLAUDE.md issues by running a diagnostic audit. This evaluates the file structure, frontmatter correctness, and actionability to produce a detailed report with severity ratings and specific fixes.

You need to check CLAUDE.md for staleness and anti-patterns when an AI assistant is struggling to understand project context or follow instructions. Diagnosing these files ensures the AI has accurate, efficient, and up-to-date information.