Check your CLAUDE.md or AGENTS.md before it costs you.
Fourteen checks over the instructions file your coding agent reads at the start of every session: contradictions, dead weight, machine-specific paths, and secrets that should never have been committed.
Runs in your browser. Whatever you drop or paste here is read by JavaScript in this tab and never sent anywhere. There is no upload endpoint on this site to send it to.
Works on CLAUDE.md, AGENTS.md, .cursorrules and anything else shaped like standing instructions for a coding agent.
- Token cost of the file, since it is paid every session
- Contradictions — the same subject told always and never
- Credentials pasted in as literals
- Absolute paths that only exist on one machine
- Instructions that describe the default rather than changing it
- Emphasis inflation — everything marked critical
- Lines that appear twice
- TODO, FIXME and unfilled placeholders
- Missing build and test commands
- Oversized code blocks and pasted directory trees
- Heading depth and walls of text
None of these are rules you have to follow. They are patterns that are usually worth a second look, each reported with its line number so you can judge it against the actual line. If you want the raw token figure for a file on its own, the token counter does that without the opinions.
Common questions
What is a CLAUDE.md file?
A Markdown file at the root of a repository holding standing instructions for a coding agent — how to build and test, what conventions to follow, what to leave alone. Claude Code reads it at the start of a session. AGENTS.md is the same idea under a tool-neutral name, and .cursorrules is Cursor's equivalent.
Why does the length of the file matter?
It is prepended to your context every session, so its cost is paid on every request, not once. A file that has grown to several thousand tokens is spending real budget on instructions the model may not need — and the longer it gets, the more likely two lines in it contradict each other.
What is wrong with a vague instruction?
Nothing, except that it does no work. "Write good code" and "be careful" describe the default rather than changing it. The instructions that earn their place are the ones a competent stranger could not have guessed: which command runs the tests, which directory is generated, which pattern this codebase deliberately does not use.
Does the linter send my file anywhere?
No. Every check runs in this tab. That matters more here than on the other tools — instructions files are exactly where a stray API key or an internal hostname tends to end up, which is also why one of the checks looks for them.
Are the findings rules I have to follow?
No. They are patterns that are usually worth a second look, each with the line number so you can judge it yourself. A finding you disagree with after reading the line is a finding you should ignore.
This page checks the one file you paste into it. It cannot see the rest of your Mac — a browser is not allowed to, which is the whole reason this page is safe to use. Inventory is the desktop app that does: it indexes every conversation Cursor, Claude Code, Zed, Codex, Kiro and Antigravity have had on your machine and searches all of them at once, privately, from ⌘⇧Space. $19.99 once, free for 7 days, no account.