Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when creating or maintaining UI design systems, design tokens, component documentation, responsive calculations, quality checks, or design-dev handoff materials.
.claude/skills/majiayu000-ui-design-system/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -80% | 0% |
Professional toolkit for creating and maintaining scalable design systems.
Generates complete design system tokens from brand colors.
Usage: python scripts/design_token_generator.py [brand_color] [style] [format]
Features:
Before reporting completion, validate generated tokens or handoff artifacts against the requested brand color, output format, accessibility constraints, and the target implementation surface. For script usage, run the command with representative input and inspect the emitted JSON/CSS/SCSS for valid syntax and preserved brand values.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 26,077 | 10,226 | -61% | 1 | 1 | 0% | 5,770 | 498 | -91% | 0 | 0 | — |
case-02 | fail→fail | 29,144 | 5,415 | -81% | 1 | 1 | 0% | 6,199 | 501 | -92% | 0 | 0 | — |
case-03 | fail→fail | 17,098 | 5,691 | -67% | 1 | 1 | 0% | 3,247 | 494 | -85% | 0 | 0 | — |
case-04 | fail→fail | 18,501 | 7,333 | -60% | 1 | 1 | 0% | 3,919 | 777 | -80% | 0 | 0 | — |
case-05 | fail→fail | 22,210 | 28,972 | +30% | 1 | 1 | 0% | 3,969 | 6,198 | +56% | 0 | 0 | — |
case-06 | fail→fail | 24,298 | 27,653 | +14% | 1 | 1 | 0% | 5,267 | 5,847 | +11% | 0 | 0 | — |
case-07 | pass→pass | 10,929 | 4,389 | -60% | 1 | 1 | 0% | 1,576 | 928 | -41% | 0 | 0 | — |
case-08 | pass→pass | 15,012 | 7,409 | -51% | 1 | 1 | 0% | 2,472 | 1,439 | -42% | 0 | 0 | — |
case-09 | pass→pass | 9,207 | 1,613 | -82% | 1 | 1 | 0% | 1,462 | 393 | -73% | 0 | 0 | — |
case-10 | fail→pass | 4,340 | 3,269 | -25% | 1 | 1 | 0% | 541 | 628 | +16% | 0 | 0 | — |
case-11 | pass→pass | 5,835 | 3,168 | -46% | 1 | 1 | 0% | 873 | 602 | -31% | 0 | 0 | — |
case-12 | pass→pass | 8,755 | 2,327 | -73% | 1 | 1 | 0% | 1,308 | 547 | -58% | 0 | 0 | — |
case-13 | pass→pass | 9,025 | 3,041 | -66% | 1 | 1 | 0% | 1,452 | 674 | -54% | 0 | 0 | — |
case-14 | pass→pass | 7,479 | 6,110 | -18% | 1 | 1 | 0% | 1,111 | 1,182 | +6% | 0 | 0 | — |
case-15 | pass→pass | 14,172 | 2,115 | -85% | 1 | 1 | 0% | 2,084 | 506 | -76% | 0 | 0 | — |
case-16 | fail→pass | 10,464 | 2,071 | -80% | 1 | 1 | 0% | 1,657 | 571 | -66% | 0 | 0 | — |
case-17 | fail→pass | 12,212 | 5,289 | -57% | 1 | 1 | 0% | 1,747 | 1,012 | -42% | 0 | 0 | — |
case-18 | fail→pass | 7,066 | 2,694 | -62% | 1 | 1 | 0% | 951 | 626 | -34% | 0 | 0 | — |
case-19 | fail→pass | 16,982 | 2,690 | -84% | 1 | 1 | 0% | 3,071 | 623 | -80% | 0 | 0 | — |
case-20 | pass→pass | 15,384 | 13,288 | -14% | 1 | 1 | 0% | 2,696 | 2,678 | -1% | 0 | 0 | — |
case-21 | pass→pass | 10,607 | 7,543 | -29% | 1 | 1 | 0% | 1,790 | 1,519 | -15% | 0 | 0 | — |
case-22 | pass→pass | 17,025 | 15,027 | -12% | 1 | 1 | 0% | 2,960 | 2,826 | -5% | 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 18 counted toward the lift figure. The other 4 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 +23 percentage points is the difference between those two pass rates over the 18 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.