Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when testing warning-only exit code behavior.
.claude/skills/ai-creed-warn-only/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -80% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -83% | 0% |
Control fixture: exactly one warning (FM01 unknown field), exit 0.
None — the fixture is static.
None — no external dependencies.
A findings list with a single FM01 warning.
Given the input shakespii lint tests/fixtures/warn-only, the expected output is one warning and exit code 0.
Adding a second unknown field — the CLI test locks the count at one.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 3,325 | 2,255 | -32% | 1 | 1 | 0% | 447 | 512 | +15% | 0 | 0 | — |
case-02 | fail→pass | 2,708 | 5,100 | +88% | 1 | 1 | 0% | 436 | 556 | +28% | 0 | 0 | — |
case-03 | fail→pass | 4,749 | 2,697 | -43% | 1 | 1 | 0% | 782 | 591 | -24% | 0 | 0 | — |
case-04 | pass→pass | 5,247 | 2,106 | -60% | 1 | 1 | 0% | 843 | 490 | -42% | 0 | 0 | — |
case-05 | pass→pass | 7,871 | 3,284 | -58% | 1 | 1 | 0% | 1,363 | 663 | -51% | 0 | 0 | — |
case-06 | pass→pass | 7,248 | 2,817 | -61% | 1 | 1 | 0% | 1,245 | 645 | -48% | 0 | 0 | — |
case-07 | fail→pass | 9,599 | 1,289 | -87% | 1 | 1 | 0% | 1,646 | 332 | -80% | 0 | 0 | — |
case-08 | pass→pass | 23,429 | 1,478 | -94% | 1 | 1 | 0% | 4,381 | 349 | -92% | 0 | 0 | — |
case-09 | pass→pass | 4,999 | 2,200 | -56% | 1 | 1 | 0% | 823 | 486 | -41% | 0 | 0 | — |
case-10 | pass→pass | 5,974 | 1,791 | -70% | 1 | 1 | 0% | 1,099 | 418 | -62% | 0 | 0 | — |
case-11 | fail→pass | 15,256 | 1,729 | -89% | 1 | 1 | 0% | 2,630 | 454 | -83% | 0 | 0 | — |
case-12 | fail→pass | 8,647 | 2,374 | -73% | 1 | 1 | 0% | 1,508 | 537 | -64% | 0 | 0 | — |
case-13 | pass→pass | 18,559 | 2,400 | -87% | 1 | 1 | 0% | 1,424 | 604 | -58% | 0 | 0 | — |
case-14 | pass→pass | 10,691 | 3,314 | -69% | 1 | 1 | 0% | 1,658 | 713 | -57% | 0 | 0 | — |
case-15 | fail→pass | 10,871 | 5,407 | -50% | 1 | 1 | 0% | 2,023 | 1,041 | -49% | 0 | 0 | — |
case-16 | pass→pass | 10,190 | 1,982 | -81% | 1 | 1 | 0% | 1,792 | 499 | -72% | 0 | 0 | — |
case-17 | fail→pass | 10,536 | 2,484 | -76% | 1 | 1 | 0% | 2,298 | 539 | -77% | 0 | 0 | — |
case-18 | fail→pass | 16,414 | 3,211 | -80% | 1 | 1 | 0% | 2,456 | 731 | -70% | 0 | 0 | — |
case-19 | fail→pass | 9,632 | 3,092 | -68% | 1 | 1 | 0% | 1,627 | 691 | -58% | 0 | 0 | — |
case-20 | pass→pass | 7,789 | 1,522 | -80% | 1 | 1 | 0% | 1,486 | 351 | -76% | 0 | 0 | — |
case-21 | fail→pass | 9,579 | 2,402 | -75% | 1 | 1 | 0% | 1,645 | 436 | -73% | 0 | 0 | — |
case-22 | pass→pass | 8,486 | 3,828 | -55% | 1 | 1 | 0% | 1,246 | 813 | -35% | 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. The headline lift of +50 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.