RuleReceipt
Check which CLAUDE.md rules your AI agent actually followed
2 followers
Check which CLAUDE.md rules your AI agent actually followed
2 followers
RuleReceipt checks a Claude Code session against your own rules and shows, per rule, what was followed, what wasn't, and what it couldn't tell — with the quoted evidence. Runs locally. No account. Nothing leaves your machine unless you choose to send it.
Free
Launch Team / Built With
CotypistType up to 50% less in every Mac app. Free to try, and it runs fully on your Mac.
Promoted
Hi Product Hunt 👋
My rules file said "never push without asking". Claude pushed to main anyway.
So I built RuleReceipt. It checks your real sessions against the rules you already wrote: followed, broken, or couldn't tell, with the exact line as proof.
Runs locally. No account. Nothing leaves your machine unless you choose to send it.
It only says "broken" with hard evidence. False-accusation ceiling: 0.8% (method at rulereceipt.dev/accuracy).
Verdict looks wrong? Every verdict has an ID. Run:
It builds a short report with secrets masked, shows it to you, then you choose: send it as a public GitHub issue, or privately to hello@rulereceipt.dev. Nothing is sent without your OK. I'll fix it the same day.
Claude Code today, Codex in testing.
No Claude Code history on your machine? You can still try it in under a minute:
🌐 In your browser, nothing to install: rulereceipt.dev/check (paste a rules file, drop a session file; nothing leaves your browser)
💻 Or in a terminal: npx rulereceipt demo
If something's missing or confusing, add it with "Suggest what ships next". I'll ship the small ones today 🚢
Honest question: is the proof it shows useful to you?