revise-claude-md

To audit a CLAUDE.md file, evaluate its documentation against quality criteria for commands, architecture, and actionability. This process discovers outdated or incomplete project guidance and produces a focused report proposing updates to preserve useful codebase context.

Claude Code may misunderstand architecture because the CLAUDE.md project memory documentation has become stale or incomplete. A structured repository audit evaluates codebase context against currency and pattern criteria to identify outdated instructions reducing AI comprehension.

The best way to update stale AI coding instructions is running a controlled documentation review workflow. This evaluates existing CLAUDE.md files for actionability, proposes targeted improvements that remove unnecessary information, and preserves useful codebase context.

Yes, you can use a structured documentation review to fix incomplete project memory. The audit discovers CLAUDE.md files, evaluates them against quality criteria, and proposes focused updates that enhance actionability while removing unnecessary information.

No, the CLAUDE.md update workflow does not require manual file discovery. The audit process autonomously discovers CLAUDE.md files across the repository and evaluates them against quality criteria before proposing controlled documentation updates.

You should not use an automated CLAUDE.md audit when your repository lacks existing CLAUDE.md files or when project memory documentation has not yet been established. The audit requires existing files to evaluate against quality criteria and propose targeted improvements.