Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always
.claude/skills/ibrahim-3d-verification-before-completion/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✓→✗ | ▼ Worse | -24% | 0% |
| case-12 | ✓→✗ | ▼ Worse | 21% | 0% |
| case-13 | ✓→✗ | ▼ Worse | 0% | 0% |
| case-06 | ✓→✓ | = Same ✓ | 6% | 0% |
| case-04 | ✓→✓ | = Same ✓ | -13% | 0% |
Claiming work is complete without verification is dishonesty, not efficiency.
Core principle: Evidence before claims, always.
Violating the letter of this rule is violating the spirit of this rule.
NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCEIf you haven't run the verification command in this message, you cannot claim it passes.
BEFORE claiming any status or expressing satisfaction:
1. IDENTIFY: What command proves this claim?
2. RUN: Execute the FULL command (fresh, complete)
3. read_file: Full output, check exit code, count failures
4. VERIFY: Does output confirm the claim?
- If NO: State actual status with evidence
- If YES: State claim WITH evidence
5. ONLY THEN: Make the claim
Skip any step = lying, not verifying| Claim | Requires | Not Sufficient | |-------|----------|----------------| | Tests pass | Test command output: 0 failures | Previous run, "should pass" | | Linter clean | Linter output: 0 errors | Partial check, extrapolation | | Build succeeds | Build command: exit 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 | | Agent completed | VCS diff shows changes | Agent reports "success" | | Requirements met | Line-by-line checklist | Tests passing |
| Excuse | Reality | |--------|---------| | "Should work now" | RUN the verification | | "I'm confident" | Confidence ≠ evidence | | "Just this once" | No exceptions | | "Linter passed" | Linter ≠ compiler | | "Agent said success" | Verify independently | | "I'm tired" | Exhaustion ≠ excuse | | "Partial check is enough" | Partial proves nothing | | "Different words so rule doesn't apply" | Spirit over letter |
Tests:
✅ [Run test command] [See: 34/34 pass] "All tests pass"
❌ "Should pass now" / "Looks correct"Regression tests (TDD Red-Green):
✅ write_file → Run (pass) → Revert fix → Run (MUST FAIL) → Restore → Run (pass)
❌ "I've written a regression test" (without red-green verification)Build:
✅ [Run build] [See: exit 0] "Build passes"
❌ "Linter passed" (linter doesn't check compilation)Requirements:
✅ Re-read_file plan → Create checklist → Verify each → Report gaps or completion
❌ "Tests pass, phase complete"Agent delegation:
✅ Agent reports success → Check VCS diff → Verify changes → Report actual state
❌ Trust agent reportFrom 24 failure memories:
ALWAYS before:
Rule applies to:
No shortcuts for verification.
Run the command. read_file the output. THEN claim the result.
This is non-negotiable.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | fail→fail | 8,038 | 13,407 | +67% | 1 | 1 | 0% | 1,498 | 2,918 | +95% | 0 | 0 | — |
case-22 | fail→fail | 6,363 | 15,452 | +143% | 1 | 1 | 0% | 1,003 | 3,180 | +217% | 0 | 0 | — |
case-06 | pass→pass | 12,371 | 7,594 | -39% | 1 | 1 | 0% | 2,255 | 2,397 | +6% | 0 | 0 | — |
case-01 | fail→fail | 6,308 | 6,794 | +8% | 1 | 1 | 0% | 177 | 1,225 | +592% | 0 | 0 | — |
case-02 | fail→fail | 4,563 | 4,973 | +9% | 1 | 1 | 0% | 189 | 1,217 | +544% | 0 | 0 | — |
case-03 | fail→fail | 5,167 | 5,667 | +10% | 1 | 1 | 0% | 314 | 1,442 | +359% | 0 | 0 | — |
case-04 | pass→pass | 14,602 | 7,661 | -48% | 1 | 1 | 0% | 2,596 | 2,266 | -13% | 0 | 0 | — |
case-05 | pass→pass | 7,659 | 5,921 | -23% | 1 | 1 | 0% | 1,449 | 2,072 | +43% | 0 | 0 | — |
case-07 | fail→fail | 5,771 | 13,430 | +133% | 1 | 1 | 0% | 831 | 1,318 | +59% | 0 | 0 | — |
case-08 | fail→fail | 3,252 | 9,455 | +191% | 1 | 1 | 0% | 537 | 1,682 | +213% | 0 | 0 | — |
case-09 | fail→fail | 6,308 | 3,619 | -43% | 1 | 1 | 0% | 967 | 1,407 | +46% | 0 | 0 | — |
case-10 | fail→fail | 7,457 | 7,782 | +4% | 1 | 1 | 0% | 1,279 | 1,488 | +16% | 0 | 0 | — |
case-11 | pass→fail | 10,359 | 4,697 | -55% | 1 | 1 | 0% | 1,688 | 1,291 | -24% | 0 | 0 | — |
case-12 | pass→fail | 8,164 | 3,952 | -52% | 1 | 1 | 0% | 1,288 | 1,558 | +21% | 0 | 0 | — |
case-13 | pass→fail | 7,891 | 5,734 | -27% | 1 | 1 | 0% | 1,271 | 1,266 | -0% | 0 | 0 | — |
case-14 | fail→fail | 4,392 | 3,971 | -10% | 1 | 1 | 0% | 749 | 1,236 | +65% | 0 | 0 | — |
case-15 | fail→fail | 10,225 | 10,288 | +1% | 1 | 1 | 0% | 1,807 | 2,432 | +35% | 0 | 0 | — |
case-16 | fail→fail | 8,190 | 3,968 | -52% | 1 | 1 | 0% | 1,499 | 1,585 | +6% | 0 | 0 | — |
case-17 | fail→fail | 9,022 | 4,559 | -49% | 1 | 1 | 0% | 1,451 | 1,267 | -13% | 0 | 0 | — |
case-18 | fail→fail | 10,471 | 7,705 | -26% | 1 | 1 | 0% | 1,784 | 2,267 | +27% | 0 | 0 | — |
case-19 | fail→fail | 8,901 | 6,009 | -32% | 1 | 1 | 0% | 1,431 | 1,322 | -8% | 0 | 0 | — |
case-20 | fail→fail | 6,963 | 6,939 | -0% | 1 | 1 | 0% | 1,177 | 1,379 | +17% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 11 counted toward the lift figure. The other 11 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -14 percentage points is the difference between those two pass rates over the 11 comparable cases. 5 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.