Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create design systems with tokens, components, and documentation. Use when building design systems, creating component libraries, or establishing design tokens.
.claude/skills/onewave-ai-design-system-generator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -62% | 0% |
Generate a token-driven design system: tokens, base components, build config, and documentation.
references/design-tokens.md — color, typography, spacing, shadow, and radius token files plus CSS variable setup.references/components.md — Button, Input, Card implementations and the component barrel index.references/config-and-utilities.md — Tailwind config and the cn class-merge helper.tokens/ with colors, typography, spacing, and effects (shadows, radii) per references/design-tokens.md. Replace the reference values with the project brand palette.:root and .dark custom properties in globals.css using the variable block in references/design-tokens.md.cn helper and extend the Tailwind theme from the token files following references/config-and-utilities.md.references/components.md, then extend the set as the product requires.components/ui/index.ts (see references/components.md).| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 27,574 | 40,210 | +46% | 1 | 1 | 0% | 6,194 | 6,513 | +5% | 0 | 0 | — |
case-02 | fail→pass | 24,084 | 34,911 | +45% | 1 | 1 | 0% | 5,239 | 6,509 | +24% | 0 | 0 | — |
case-03 | fail→pass | 29,085 | 26,346 | -9% | 1 | 1 | 0% | 6,187 | 6,506 | +5% | 0 | 0 | — |
case-04 | pass→pass | 12,762 | 12,612 | -1% | 1 | 1 | 0% | 1,870 | 2,495 | +33% | 0 | 0 | — |
case-05 | fail→pass | 12,480 | 9,854 | -21% | 1 | 1 | 0% | 2,454 | 2,205 | -10% | 0 | 0 | — |
case-06 | fail→pass | 13,209 | 2,674 | -80% | 1 | 1 | 0% | 2,024 | 772 | -62% | 0 | 0 | — |
case-07 | pass→pass | 11,172 | 5,460 | -51% | 1 | 1 | 0% | 1,877 | 1,254 | -33% | 0 | 0 | — |
case-08 | pass→pass | 11,489 | 6,837 | -40% | 1 | 1 | 0% | 1,759 | 1,434 | -18% | 0 | 0 | — |
case-09 | pass→pass | 11,186 | 15,401 | +38% | 1 | 1 | 0% | 1,817 | 3,491 | +92% | 0 | 0 | — |
case-10 | pass→pass | 12,687 | 8,320 | -34% | 1 | 1 | 0% | 2,373 | 1,856 | -22% | 0 | 0 | — |
case-11 | pass→pass | 12,945 | 15,468 | +19% | 1 | 1 | 0% | 2,190 | 2,837 | +30% | 0 | 0 | — |
case-12 | pass→pass | 13,384 | 5,788 | -57% | 1 | 1 | 0% | 2,322 | 1,318 | -43% | 0 | 0 | — |
case-13 | pass→pass | 8,716 | 6,853 | -21% | 1 | 1 | 0% | 1,576 | 1,697 | +8% | 0 | 0 | — |
case-14 | pass→pass | 13,723 | 8,024 | -42% | 1 | 1 | 0% | 2,154 | 1,682 | -22% | 0 | 0 | — |
case-15 | fail→fail | 12,989 | 6,737 | -48% | 1 | 1 | 0% | 1,914 | 1,588 | -17% | 0 | 0 | — |
case-20 | fail→fail | 6,729 | 3,709 | -45% | 1 | 1 | 0% | 1,006 | 637 | -37% | 0 | 0 | — |
case-16 | pass→pass | 14,829 | 8,735 | -41% | 1 | 1 | 0% | 2,082 | 1,726 | -17% | 0 | 0 | — |
case-17 | pass→pass | 12,845 | 6,397 | -50% | 1 | 1 | 0% | 1,728 | 1,253 | -27% | 0 | 0 | — |
case-18 | fail→pass | 13,023 | 12,195 | -6% | 1 | 1 | 0% | 1,986 | 2,344 | +18% | 0 | 0 | — |
case-19 | pass→pass | 15,195 | 10,809 | -29% | 1 | 1 | 0% | 2,443 | 2,110 | -14% | 0 | 0 | — |
case-21 | pass→pass | 15,272 | 16,896 | +11% | 1 | 1 | 0% | 2,890 | 3,381 | +17% | 0 | 0 | — |
case-22 | pass→pass | 17,881 | 17,032 | -5% | 1 | 1 | 0% | 2,884 | 3,269 | +13% | 0 | 0 | — |
case-23 | pass→pass | 16,302 | 17,425 | +7% | 1 | 1 | 0% | 3,511 | 3,850 | +10% | 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. 23 cases were attempted. The headline lift of +26 percentage points is the difference between those two pass rates over the 23 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.