Trustbut — AI skill for Claude Code

---

trustbut is a local command-line tool that produces a **receipt** for any change an AI coding agent makes. It reruns your tests from scratch, checks whether the agent's new tests actually prove anything, and flags skipped tests, deleted assertions and silenced type checks. Then it gives you a **PASS / WARN / FAIL** verdict you can read in 30 seconds.

It works with **any agent** (Claude Code, Codex, Cursor, Gemini CLI, OpenCode, Copilot) because it only needs git and your test runner. No LLM, no server, no account. Your code never leaves your machine.

$ npx github:isAlmogK/trustbut verify

## 🧾 Receipt: WARN

`3f2a91c` → `8be04d1` · runner: vitest

**Tests (clean checkout):** 41 passed · 0 failed · 1 skipped
**New/changed tests:** 1 prove the change · 1 pass without it

### Findings
- ⚠ **Test passes without the change** · `test/auth.test.ts`: rejects expired token
- ⚠ **Test skipped** · `test/auth.test.ts:88`: it.skip('handles clock skew', () => {
- ⚠ **cast to any** · `src/auth/session.ts:41`: const claims = decoded as any;

### Review these first
1. `src/auth/session.ts`: auth/permissions, has findings
2. `package.json`: dependencies

The output is Markdown