Free tool
CLAUDE.md Auditor
Score your CLAUDE.md on actionability, hard rules, signal density, completeness, and agent ergonomics.
Get started-
Scores actionability, hard rules, signal density, completeness, and agent ergonomics separately — not one vague number.
-
Every finding references something concrete in your actual CLAUDE.md, with a specific fix — for hard-rules and completeness gaps, the exact line to add.
-
Catches the failure mode that actually bites: buried or missing hard rules an agent would miss on a real task.
Frequently asked questions
This tool is powered by Claude (Anthropic).
One that a coding agent can read once and behave correctly from: concrete build/test/lint commands, prominent and unambiguous hard rules, and no filler that wastes the context window it occupies every session.
Every run — including the file you paste — is logged the same way as any other Tool Platform run, for abuse prevention and diagnostics. It is not added to any list, shared, or used for anything beyond that. See our privacy policy for details.
It scores substance, not prose: whether hard rules are stated and prominent, whether instructions are concrete enough for an agent to execute, and whether the file is bloated with generic advice a fresh agent does not need.