Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A well-formed skill used as a happy-path fixture for parsing.
.claude/skills/rwrife-valid-skill/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-16 | ✓→✗ | ▼ Worse | -16% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -30% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -31% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | -14% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | -41% | 0% |
This is the body. It has a heading and a paragraph so token/secret rules have something to chew on later.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 1,923 | 3,439 | +79% | 1 | 1 | 0% | 311 | 615 | +98% | 0 | 0 | — |
case-02 | fail→fail | 14,520 | 12,914 | -11% | 1 | 1 | 0% | 1,275 | 793 | -38% | 0 | 0 | — |
case-03 | fail→fail | 23,805 | 8,172 | -66% | 1 | 1 | 0% | 1,299 | 1,701 | +31% | 0 | 0 | — |
case-04 | pass→pass | 8,760 | 6,504 | -26% | 1 | 1 | 0% | 1,661 | 1,155 | -30% | 0 | 0 | — |
case-05 | pass→pass | 13,266 | 8,081 | -39% | 1 | 1 | 0% | 2,608 | 1,796 | -31% | 0 | 0 | — |
case-06 | pass→pass | 6,520 | 4,894 | -25% | 1 | 1 | 0% | 1,210 | 1,046 | -14% | 0 | 0 | — |
case-07 | pass→pass | 17,796 | 10,644 | -40% | 1 | 1 | 0% | 3,359 | 1,975 | -41% | 0 | 0 | — |
case-08 | fail→fail | 19,258 | 15,942 | -17% | 1 | 1 | 0% | 1,621 | 1,461 | -10% | 0 | 0 | — |
case-09 | fail→fail | 18,950 | 7,574 | -60% | 1 | 1 | 0% | 2,060 | 1,245 | -40% | 0 | 0 | — |
case-10 | pass→pass | 11,911 | 15,058 | +26% | 1 | 1 | 0% | 2,099 | 2,075 | -1% | 0 | 0 | — |
case-11 | fail→fail | 14,361 | 13,345 | -7% | 1 | 1 | 0% | 1,698 | 1,543 | -9% | 0 | 0 | — |
case-12 | pass→pass | 11,839 | 9,804 | -17% | 1 | 1 | 0% | 2,322 | 2,068 | -11% | 0 | 0 | — |
case-13 | pass→pass | 5,331 | 4,708 | -12% | 1 | 1 | 0% | 885 | 858 | -3% | 0 | 0 | — |
case-14 | pass→pass | 16,390 | 8,522 | -48% | 1 | 1 | 0% | 2,561 | 1,657 | -35% | 0 | 0 | — |
case-15 | pass→pass | 17,360 | 18,936 | +9% | 1 | 1 | 0% | 1,933 | 1,867 | -3% | 0 | 0 | — |
case-16 | pass→fail | 12,788 | 13,188 | +3% | 1 | 1 | 0% | 2,537 | 2,130 | -16% | 0 | 0 | — |
case-17 | fail→fail | 21,978 | 14,757 | -33% | 1 | 1 | 0% | 2,013 | 963 | -52% | 0 | 0 | — |
case-18 | fail→fail | 7,356 | 5,332 | -28% | 1 | 1 | 0% | 1,363 | 962 | -29% | 0 | 0 | — |
case-19 | pass→pass | 9,957 | 7,494 | -25% | 1 | 1 | 0% | 2,148 | 1,448 | -33% | 0 | 0 | — |
case-20 | pass→pass | 2,735 | 2,793 | +2% | 1 | 1 | 0% | 329 | 269 | -18% | 0 | 0 | — |
case-21 | fail→fail | 2,708 | 2,610 | -4% | 1 | 1 | 0% | 448 | 451 | +1% | 0 | 0 | — |
case-22 | fail→fail | 5,631 | 2,778 | -51% | 1 | 1 | 0% | 928 | 409 | -56% | 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 21 counted toward the lift figure. The other 1 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 -100 percentage points is the difference between those two pass rates over the 21 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Other measured skills in the registry, with their headline benchmark lift.