Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Map this token system to or from any external design system (Material Design 3, Apple HIG, Fluent, Carbon, Ant, shadcn/ui, Radix, Chakra, Mantine, Bootstrap…) — adopt their look, build on their stack, or migrate between systems. Use when the user mentions interop, migration, or a specific design-system/component-library bridge.
.claude/skills/plugin87-migrate-design-system/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 95% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -41% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -51% | 0% |
Bridge to or from external design systems via a role-based crosswalk.
design-systems/interop-protocol.md (Crosswalk Method, the three directions, headless-vs-styled guidance, verification).design-systems/crosswalk.md for Material 3 / Apple HIG / Fluent 2 / Carbon / shadcn/ui / Radix. For others, derive a mapping with the Crosswalk Method (map by role/intent across 6 axes: color roles, type scale, spacing unit, radius, elevation, motion).semantic.*.scripts/contrast.py / a11y-audit); confirm all 8 states + dark mode survive the mapping.A crosswalk table (our token → their token → value note), a bridge plan if migrating, and verified token overrides. Render via design-code.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | pass→pass | 19,170 | 31,101 | +62% | 1 | 1 | 0% | 3,072 | 6,469 | +111% | 0 | 0 | — |
case-01 | fail→fail | 29,251 | 26,359 | -10% | 1 | 1 | 0% | 6,074 | 5,981 | -2% | 0 | 0 | — |
case-02 | pass→pass | 28,782 | 30,780 | +7% | 1 | 1 | 0% | 5,352 | 6,487 | +21% | 0 | 0 | — |
case-03 | fail→fail | 32,282 | 25,291 | -22% | 1 | 1 | 0% | 6,196 | 5,382 | -13% | 0 | 0 | — |
case-04 | fail→pass | 10,759 | 6,679 | -38% | 1 | 1 | 0% | 1,969 | 1,470 | -25% | 0 | 0 | — |
case-05 | fail→pass | 18,759 | 14,818 | -21% | 1 | 1 | 0% | 2,989 | 2,549 | -15% | 0 | 0 | — |
case-06 | pass→pass | 25,261 | 29,081 | +15% | 1 | 1 | 0% | 4,478 | 6,479 | +45% | 0 | 0 | — |
case-08 | fail→pass | 32,166 | 12,568 | -61% | 1 | 1 | 0% | 1,159 | 2,261 | +95% | 0 | 0 | — |
case-09 | fail→pass | 10,053 | 4,679 | -53% | 1 | 1 | 0% | 1,732 | 1,018 | -41% | 0 | 0 | — |
case-10 | fail→pass | 9,191 | 2,600 | -72% | 1 | 1 | 0% | 1,352 | 668 | -51% | 0 | 0 | — |
case-11 | fail→pass | 15,031 | 14,694 | -2% | 1 | 1 | 0% | 2,385 | 2,703 | +13% | 0 | 0 | — |
case-21 | pass→pass | 10,462 | 13,736 | +31% | 1 | 1 | 0% | 1,718 | 2,457 | +43% | 0 | 0 | — |
case-12 | fail→fail | 14,774 | 18,247 | +24% | 1 | 1 | 0% | 2,378 | 4,054 | +70% | 0 | 0 | — |
case-13 | fail→fail | 17,090 | 8,134 | -52% | 1 | 1 | 0% | 2,606 | 1,561 | -40% | 0 | 0 | — |
case-14 | fail→pass | 13,472 | 10,563 | -22% | 1 | 1 | 0% | 2,166 | 2,235 | +3% | 0 | 0 | — |
case-15 | fail→pass | 9,463 | 3,902 | -59% | 1 | 1 | 0% | 1,463 | 882 | -40% | 0 | 0 | — |
case-16 | pass→pass | 43,197 | 11,579 | -73% | 1 | 1 | 0% | 1,994 | 2,163 | +8% | 0 | 0 | — |
case-17 | fail→pass | 8,412 | 3,282 | -61% | 1 | 1 | 0% | 1,211 | 762 | -37% | 0 | 0 | — |
case-18 | fail→pass | 22,174 | 3,626 | -84% | 1 | 1 | 0% | 1,182 | 868 | -27% | 0 | 0 | — |
case-19 | pass→pass | 8,197 | 15,890 | +94% | 1 | 1 | 0% | 1,244 | 3,304 | +166% | 0 | 0 | — |
case-20 | pass→pass | 11,683 | 13,376 | +14% | 1 | 1 | 0% | 2,401 | 2,865 | +19% | 0 | 0 | — |
case-22 | fail→fail | 11,929 | 19,245 | +61% | 1 | 1 | 0% | 2,165 | 4,673 | +116% | 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 20 counted toward the lift figure. The other 2 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 +45 percentage points is the difference between those two pass rates over the 20 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.