Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create or review accessible web, desktop, iOS, Android, and cross-platform UI/UX. Use for layouts, themes, components, design systems, DESIGN.md tokens, responsive behavior, dark mode, motion, and WCAG validation.
.claude/skills/github-gem-design-md-guidelines/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 65% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 162% | 0% |
Platform.select or a framework adapter; never duplicate whole designs for superficial differences.StyleSheet.create or framework theme. Never use inline styles for static values.DESIGN.md complianceUse Google DESIGN.md alpha format:
version, name, description, colors, typography, rounded, spacing, components.## Overview, ## Colors, ## Typography, ## Layout, ## Elevation & Depth, ## Shapes, ## Components, ## Do's and Don'ts.Every YAML components: value MUST be a {token.ref}—never an inline color, spacing, dimension, or other raw value. Run npx @google/design.md lint DESIGN.md when available.
expo-font, react-native-google-fonts, or embedded assets).Allowed when justified: asymmetric grids, controlled overlap/negative margins/z-index, bento or diagonal flow, full-bleed media with contained content, varied-height mobile lists, snap scrolling, reachable floating controls, and safe-area-aware bottom sheets. Keep them responsive, readable, keyboard accessible, and free of unintended horizontal overflow.
Apply in order:
StyleSheet.create (React Native) or framework theme (Flutter), using tokens.Platform.select only for genuine differences such as shadows, fonts, or spacing.DESIGN.md tokens and StyleSheet.create; no hardcoded or static inline styles.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 49,819 | 47,056 | -6% | 1 | 1 | 0% | 8,256 | 9,924 | +20% | 0 | 0 | — |
case-02 | fail→pass | 45,409 | 45,119 | -1% | 1 | 1 | 0% | 7,988 | 9,920 | +24% | 0 | 0 | — |
case-03 | fail→fail | 56,079 | 48,336 | -14% | 1 | 1 | 0% | 8,056 | 9,917 | +23% | 0 | 0 | — |
case-04 | pass→pass | 19,478 | 27,623 | +42% | 1 | 1 | 0% | 2,157 | 5,646 | +162% | 0 | 0 | — |
case-05 | pass→pass | 17,812 | 14,403 | -19% | 1 | 1 | 0% | 1,920 | 2,989 | +56% | 0 | 0 | — |
case-06 | pass→pass | 19,659 | 35,019 | +78% | 1 | 1 | 0% | 2,395 | 7,152 | +199% | 0 | 0 | — |
case-07 | pass→pass | 18,031 | 10,170 | -44% | 1 | 1 | 0% | 1,947 | 3,209 | +65% | 0 | 0 | — |
case-08 | pass→pass | 21,212 | 23,981 | +13% | 1 | 1 | 0% | 2,384 | 4,393 | +84% | 0 | 0 | — |
case-09 | pass→pass | 16,822 | 13,508 | -20% | 1 | 1 | 0% | 2,008 | 3,184 | +59% | 0 | 0 | — |
case-10 | pass→pass | 22,686 | 26,192 | +15% | 1 | 1 | 0% | 2,651 | 5,180 | +95% | 0 | 0 | — |
case-11 | pass→pass | 21,158 | 22,687 | +7% | 1 | 1 | 0% | 2,528 | 4,924 | +95% | 0 | 0 | — |
case-12 | fail→pass | 15,466 | 11,936 | -23% | 1 | 1 | 0% | 1,686 | 2,779 | +65% | 0 | 0 | — |
case-13 | pass→pass | 6,134 | 8,275 | +35% | 1 | 1 | 0% | 744 | 2,113 | +184% | 0 | 0 | — |
case-14 | pass→pass | 16,066 | 17,920 | +12% | 1 | 1 | 0% | 2,395 | 3,574 | +49% | 0 | 0 | — |
case-15 | fail→pass | 15,769 | 12,964 | -18% | 1 | 1 | 0% | 1,639 | 2,851 | +74% | 0 | 0 | — |
case-16 | fail→fail | 18,162 | 37,409 | +106% | 1 | 1 | 0% | 2,623 | 6,406 | +144% | 0 | 0 | — |
case-17 | fail→pass | 21,800 | 10,916 | -50% | 1 | 1 | 0% | 2,526 | 2,738 | +8% | 0 | 0 | — |
case-18 | pass→pass | 17,556 | 29,796 | +70% | 1 | 1 | 0% | 2,405 | 5,417 | +125% | 0 | 0 | — |
case-19 | pass→pass | 17,318 | 29,607 | +71% | 1 | 1 | 0% | 2,218 | 5,425 | +145% | 0 | 0 | — |
case-20 | pass→pass | 19,895 | 20,268 | +2% | 1 | 1 | 0% | 2,628 | 4,536 | +73% | 0 | 0 | — |
case-21 | pass→pass | 15,495 | 24,084 | +55% | 1 | 1 | 0% | 2,886 | 5,459 | +89% | 0 | 0 | — |
case-22 | pass→pass | 30,996 | 24,598 | -21% | 1 | 1 | 0% | 5,032 | 6,254 | +24% | 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 +18 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.