Saddss Vendored from obra/superpowers under MIT. Inference triggers added to description.
Verification Before Completion
Core principle: Evidence before claims, always.
Iron Law
NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
If you have not run the verification command in this message, you cannot claim it passes.
Gate Function
BEFORE claiming success or expressing satisfaction:
1. IDENTIFY — what command proves the claim?
2. RUN — full command, fresh, complete
3. READ — full output, exit code, failure count
4. VERIFY — output supports the claim?
5. ONLY THEN — state the claim WITH evidence
Skip any step = not verified.
Common Failures
| Claim | Requires | Not sufficient |
|---|---|---|
| Tests pass | Command output: 0 failures | Previous run, "should pass" |
| Linter clean | Linter: 0 errors | Partial file check |
| Build succeeds | Build exit 0 | Linter only |
| Bug fixed | Repro test passes | Code changed, assumed fixed |
| Benchmark improved | Rerun with numbers | Single old log |
| Agent done | VCS diff + verify | Agent said "success" |
Red Flags — STOP
- "should", "probably", "seems to"
- "Done!", "Perfect!", "Great!" before verification
- Commit/PR without running checks
- Trusting subagent reports without independent verify
Patterns
✅ [pytest …] → 42 passed → "All tests pass"
❌ "Should pass now"
✅ Revert fix → test FAILS → restore → PASS (regression test)
❌ "Added a regression test" (no red-green proof)
When to Apply
Always before: completion claims, satisfaction, commit, PR, next task, delegating to agents.
Integration
Final gate after executing-plans, tdd, simplify-code, diagnose, and benchmark runs.