Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate, extend, or audit design tokens in DTCG format with the 3-tier architecture (primitive → semantic → component). Use when the user wants a color palette, type scale, spacing/shadow/radius/motion tokens, multi-brand theming, or wants to validate token files. Covers colors, typography, spacing, shadows, borders, breakpoints, motion, gradients, opacity, blur, sizing, states, theming.
.claude/skills/plugin87-design-tokens/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -4% | 0% |
Produce and maintain DTCG ($type/$value) tokens following the project's 3-tier system.
CLAUDE.md → "Token System" + "Color/Typography/Spacing Guidelines" for the rules (4px base, Major Third scale, OKLCH palette generation, dark-mode-at-semantic-layer).tokens/ to match structure: colors.json, typography.json, spacing.json, shadows.json, borders.json, breakpoints.json, motion.json, gradients.json, opacity.json, blur.json, sizing.json, states.json, theming.json.a11y-audit skill / scripts/contrast.py.theming.json.python3 scripts/validate_tokens.py (JSON validity + alias resolution).DTCG JSON. Preserve $description on every token. Reference, never hardcode.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | pass→pass | 11,557 | 13,767 | +19% | 1 | 1 | 0% | 1,915 | 2,852 | +49% | 0 | 0 | — |
case-03 | fail→pass | 15,771 | 21,032 | +33% | 1 | 1 | 0% | 3,330 | 5,106 | +53% | 0 | 0 | — |
case-01 | fail→fail | 27,888 | 5,640 | -80% | 1 | 1 | 0% | 3,220 | 630 | -80% | 0 | 0 | — |
case-04 | pass→pass | 6,507 | 3,750 | -42% | 1 | 1 | 0% | 1,076 | 883 | -18% | 0 | 0 | — |
case-05 | pass→pass | 8,023 | 3,941 | -51% | 1 | 1 | 0% | 1,107 | 954 | -14% | 0 | 0 | — |
case-06 | pass→pass | 12,526 | 7,263 | -42% | 1 | 1 | 0% | 2,250 | 1,547 | -31% | 0 | 0 | — |
case-07 | fail→pass | 11,593 | 2,169 | -81% | 1 | 1 | 0% | 2,011 | 656 | -67% | 0 | 0 | — |
case-08 | fail→pass | 3,867 | 4,203 | +9% | 1 | 1 | 0% | 544 | 776 | +43% | 0 | 0 | — |
case-09 | fail→pass | 13,409 | 5,121 | -62% | 1 | 1 | 0% | 2,446 | 1,319 | -46% | 0 | 0 | — |
case-10 | pass→pass | 12,210 | 7,957 | -35% | 1 | 1 | 0% | 1,972 | 1,641 | -17% | 0 | 0 | — |
case-11 | pass→pass | 5,811 | 2,160 | -63% | 1 | 1 | 0% | 955 | 633 | -34% | 0 | 0 | — |
case-12 | fail→pass | 13,197 | 10,561 | -20% | 1 | 1 | 0% | 2,329 | 2,228 | -4% | 0 | 0 | — |
case-13 | pass→pass | 7,458 | 3,947 | -47% | 1 | 1 | 0% | 1,135 | 984 | -13% | 0 | 0 | — |
case-14 | fail→pass | 15,123 | 9,209 | -39% | 1 | 1 | 0% | 2,249 | 1,913 | -15% | 0 | 0 | — |
case-15 | pass→pass | 7,716 | 4,404 | -43% | 1 | 1 | 0% | 1,263 | 1,064 | -16% | 0 | 0 | — |
case-16 | pass→pass | 6,814 | 4,960 | -27% | 1 | 1 | 0% | 1,137 | 1,170 | +3% | 0 | 0 | — |
case-17 | pass→pass | 5,113 | 4,520 | -12% | 1 | 1 | 0% | 853 | 943 | +11% | 0 | 0 | — |
case-18 | fail→pass | 7,726 | 2,470 | -68% | 1 | 1 | 0% | 1,161 | 655 | -44% | 0 | 0 | — |
case-19 | fail→pass | 10,753 | 5,290 | -51% | 1 | 1 | 0% | 1,778 | 1,223 | -31% | 0 | 0 | — |
case-20 | pass→pass | 8,061 | 5,708 | -29% | 1 | 1 | 0% | 1,239 | 1,255 | +1% | 0 | 0 | — |
case-21 | pass→pass | 2,869 | 3,193 | +11% | 1 | 1 | 0% | 390 | 835 | +114% | 0 | 0 | — |
case-22 | pass→pass | 2,276 | 2,813 | +24% | 1 | 1 | 0% | 303 | 802 | +165% | 0 | 0 | — |
case-23 | pass→pass | 7,264 | 6,561 | -10% | 1 | 1 | 0% | 1,217 | 1,376 | +13% | 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, and 22 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 +35 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.