Verification Protocol
Source: Adapted from obra/superpowers:verification-before-completion
Core Principle
text
NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
Claiming work is complete without verification is dishonesty, not efficiency.
The Iron Law
Evidence before claims, always.
If you haven't run the verification command in this current context, you cannot claim it passes.
The Gate Function
BEFORE claiming any status or completion:
- •IDENTIFY: What command proves this claim?
- •RUN: Execute the FULL command (fresh, complete)
- •READ: Full output, check exit code, count failures
- •VERIFY: Does output confirm the claim?
- •If NO: State actual status with evidence
- •If YES: State claim WITH evidence
- •ONLY THEN: Make the claim
Skip any step = invalid claim
Automatic Integration
This protocol is automatically enforced in:
- •
/git:prepare-pr- Before creating PR - •
/test- Before claiming tests pass - •
/db- Before claiming migration applied - •
/review- Before claiming lint/type clean
Verification Commands by Context
| Context | Verification Command | Success Indicator |
|---|---|---|
| Tests | pnpm test | "X/X tests pass" |
| TypeScript | pnpm typecheck | "No errors found" |
| Linting | pnpm lint | "0 errors, 0 warnings" |
| Build | pnpm build | "Build succeeded" |
| Migration | pnpm db:migrate | "Migration X applied" |
| Coverage | pnpm test:coverage | "Coverage: X%" |
| E2E Tests | pnpm test:e2e | "X specs passed" |
| RLS Tests | pnpm test:rls | "X RLS tests passed" |
| Regression Test (TDD) | Write → Run (should fail) → Implement → Run (should pass) | Red-Green cycle verified |
Red Flags - STOP
If you catch yourself:
- •Using "should", "probably", "seems to"
- •Expressing satisfaction before verification ("Great!", "Perfect!", "Done!")
- •About to commit/push/PR without verification
- •Relying on partial verification
- •Thinking "just this once"
- •ANY wording implying success without having run verification
STOP. Run verification first.
Common Failures
| Claim | Requires | Not Sufficient |
|---|---|---|
| "Tests pass" | pnpm test output: 0 failures | Previous run, "should pass" |
| "Linter clean" | pnpm lint output: 0 errors/warnings | Partial check, extrapolation |
| "Build succeeds" | pnpm build exit code 0 | Linter passing, logs look good |
| "Bug fixed" | Test original symptom: passes | Code changed, assumed fixed |
| "Regression test works" | Red-green cycle verified | Test passes once |
| "Requirements met" | Line-by-line checklist | Tests passing |
Rationalization Prevention
| Excuse | Reality |
|---|---|
| "Should work now" | RUN the verification |
| "I'm confident" | Confidence ≠ evidence |
| "Just this once" | No exceptions |
| "Linter passed" | Linter ≠ compiler ≠ tests |
| "I'm tired" | Exhaustion ≠ excuse |
| "Partial check is enough" | Partial proves nothing |
| "Different words so rule doesn't apply" | Spirit over letter |
Key Patterns
Pattern: Tests
text
✅ CORRECT: [Runs: pnpm test] [Sees output: "34/34 tests passed"] "All tests pass. Output shows 34/34 tests passed with no failures." ❌ INCORRECT: "Tests should pass now" "Looks correct" "I've fixed the issue, tests will pass"
Pattern: Regression Tests (TDD Red-Green)
text
✅ CORRECT: 1. Write test 2. Run test → MUST FAIL 3. Implement fix 4. Run test → MUST PASS 5. Revert fix temporarily 6. Run test → MUST FAIL AGAIN (proves test catches the bug) 7. Restore fix 8. Run test → PASSES "Regression test verified with red-green-red-green cycle" ❌ INCORRECT: "I've written a regression test" (without verifying failure-success cycle)
Pattern: Build
text
✅ CORRECT: [Runs: pnpm build] [Sees: "Build completed successfully", exit code 0] "Build passes. Successfully compiled all packages." ❌ INCORRECT: "Linter passed, so build should work"
Pattern: Requirements
text
✅ CORRECT: [Re-reads plan] [Creates checklist of acceptance criteria] [Verifies each item] [Reports gaps OR completion with evidence] ❌ INCORRECT: "Tests pass, phase complete"
Integration with Other Skills
This Skill is automatically invoked by:
- •
git-workflow- Before commits and PRs - •
test-scaffolder- Before claiming tests complete - •
supabase-integration- Before claiming migrations applied - •
code-reviewer- Before claiming code quality checks pass
No manual trigger needed - verification happens automatically at completion checkpoints.
Why This Matters
From observed failure patterns:
- •Trust broken when claims don't match reality
- •Undefined functions shipped (would crash in production)
- •Missing requirements shipped (incomplete features)
- •Time wasted: false completion → redirect → rework cycle
- •Violates core value: "Honesty is fundamental"
When To Apply
ALWAYS before:
- •ANY variation of success/completion claims
- •ANY expression of satisfaction
- •ANY positive statement about work state
- •Committing, PR creation, task completion
- •Moving to next task
Rule applies to:
- •Exact phrases ("tests pass")
- •Paraphrases and synonyms ("all checks green")
- •Implications of success ("ready to ship")
- •ANY communication suggesting completion/correctness
The Bottom Line
No shortcuts for verification.
Run the command. Read the output. THEN claim the result.
This is non-negotiable.
Token Cost: ~180 tokens (loaded on-demand) Enforcement: Automatic integration with all Skills Source: obra/superpowers