| name | principle-prove-it-works |
| description | Apply before declaring any task or fix done. Verify the real artifact — run it, read the actual value, inspect the diff — not a self-report, proxy, or 'it compiles.' |
Prove It Works
Verify every task output by checking the real thing directly. Do not infer from proxies, self-reports, or "it compiles."
Why: Unverified work has unknown correctness. Indirect verification (file mtimes, output freshness, agent self-reports, cached screenshots) feels cheaper than direct observation. Acting on a wrong inference costs far more than checking the source.
The Iron Law: no completion claims without fresh verification evidence. If the proving command was not run in this message, the claim cannot be made — and the rule covers paraphrases, satisfaction ("Done!", "Perfect!"), and implications of success, not just the exact words.
Pattern: After completing any task, ask: "how do I prove this actually works?"
The gate, before any status claim: identify the command that proves it → run it fresh and complete → read the full output and exit code → only then state the claim, with the evidence. If the output contradicts the claim, state the actual status instead.
| Excuse | Reality |
|---|
| "Should work now" / "I'm confident" | Confidence ≠ evidence — run it |
| "Linter passed" | Linter ≠ compiler ≠ tests |
| "Agent said success" | Verify the artifact independently |
| "Partial check is enough" | Partial proves nothing |
| "Just this once" / "I'm tired" | No exceptions |
For regression tests, verify red-green: run the new test (pass) → revert the fix (test MUST fail) → restore (pass). A test never seen failing proves nothing.
Check the real thing, not a proxy:
- Check process liveness directly, not indirectly through derived state
- Read the actual value, not a cached or derived representation
- When verification fails, suspect the observation method before suspecting the system
Code and features:
- Build it (necessary but not sufficient)
- Run it and exercise the actual feature path
- Check the full chain: does data flow from input to output?
- For integrations, test the full communication path end-to-end
Delegation: trust artifacts, not self-reports.
When verifying delegated work, inspect the actual output artifact (git diff, file contents, runtime behavior), not the delegate's summary. Agents report what they intended, not always what happened.
Script the check when you can
The strongest proof is a deterministic script that re-runs the same comparison, not a one-time eyeball. Write the script, run it, and keep its output as an artifact a reviewer can re-run instead of trusting your word. A script comparing the old and new compiled output catches what a glance misses.
Prose reminders get skipped under pressure. Put the check where it can't be: an executable .claude/verify.sh in the repo root is picked up automatically by the verify-gate.sh Stop hook, which runs it before the agent is allowed to finish and blocks with the failure output if it doesn't pass. Write or extend that script instead of re-promising to verify by hand.
Keep the artifact visible for the human. Commit it only for large or complex work where the trail has to be auditable later, like a big port or migration (the show-me-your-work skill). Most work just needs it visible, not committed.
