See what's clogging your AI agent.

Free AGENTS.md, CLAUDE.md and Cursor rules token counter. Paste or drop files to weigh tokens, spot duplicates and conflicting always/never rules, and export a trimmed copy — entirely in your browser. Nothing is uploaded.

Built for Cursor, Claude Code, Codex and similar coding agents. Try the AGENTS.md / CLAUDE.md token counter, read the guides, or browse posts. Pricing: free inspector, Pro $15/mo for named projects and local history.

See it in action

examples/sample-AGENTS.md · o200k_base

A realistic AGENTS.md scanned with the same inspector logic that runs in your browser. Numbers below are the true tokenizer output — not estimates.

Total tokens
720
1 file · 3,459 chars
Heaviest section
Code review checklist
161 tokens
Issues flagged
1 + 1
duplicate · conflict

Top heavy sections

  • Code review checklist — 161 tokens
  • Always — 119 tokens
  • Never — 88 tokens
  • Stack — 73 tokens
  • Deployment — 68 tokens

What the inspector caught

Duplicate: near-identical PR review checklist (~78 redundant tokens).

Conflict: “Always force-push to main…” vs “Never use --force when pushing to main…”.

Try it with your own file