Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when a design is ready for engineering — produce an implementation-ready spec covering layout, tokens, states, responsive behavior, edge cases, motion, and accessibility.
.claude/skills/getcrew44-design-handoff/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 80% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 60% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -4% | 0% |
Produce a spec that lets the Coding Agent build the design without guessing. If it isn't specified, they'll have to invent — and they'll invent inconsistently.
--space-4 or color-primary-600, not 16px or #2563eb. If a value isn't in the system, flag that as a token gap, not a one-off.prefers-reduced-motion.A single document with the sections above. Each section uses tables or short bullets, not prose paragraphs. Tokens and component names cited verbatim. Open questions called out at the bottom, tagged with who needs to answer (engineering, product, content).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→pass | 21,216 | 34,059 | +61% | 1 | 1 | 0% | 3,434 | 6,188 | +80% | 0 | 0 | — |
case-01 | fail→fail | 23,042 | 32,972 | +43% | 1 | 1 | 0% | 4,191 | 6,775 | +62% | 0 | 0 | — |
case-02 | fail→fail | 33,301 | 32,070 | -4% | 1 | 1 | 0% | 6,200 | 6,330 | +2% | 0 | 0 | — |
case-03 | fail→fail | 32,943 | 33,316 | +1% | 1 | 1 | 0% | 6,200 | 6,772 | +9% | 0 | 0 | — |
case-04 | fail→pass | 17,007 | 24,458 | +44% | 1 | 1 | 0% | 2,994 | 4,788 | +60% | 0 | 0 | — |
case-05 | pass→fail | 36,857 | 36,399 | -1% | 1 | 1 | 0% | 6,179 | 6,751 | +9% | 0 | 0 | — |
case-06 | fail→pass | 23,362 | 27,482 | +18% | 1 | 1 | 0% | 3,811 | 5,128 | +35% | 0 | 0 | — |
case-07 | fail→pass | 26,435 | 37,022 | +40% | 1 | 1 | 0% | 4,865 | 6,740 | +39% | 0 | 0 | — |
case-09 | pass→pass | 19,028 | 32,968 | +73% | 1 | 1 | 0% | 3,316 | 5,979 | +80% | 0 | 0 | — |
case-10 | pass→pass | 24,222 | 27,186 | +12% | 1 | 1 | 0% | 3,958 | 5,108 | +29% | 0 | 0 | — |
case-11 | pass→pass | 17,250 | 25,557 | +48% | 1 | 1 | 0% | 2,938 | 4,678 | +59% | 0 | 0 | — |
case-12 | fail→pass | 12,096 | 8,315 | -31% | 1 | 1 | 0% | 1,710 | 1,645 | -4% | 0 | 0 | — |
case-21 | pass→pass | 8,409 | 10,319 | +23% | 1 | 1 | 0% | 1,494 | 2,605 | +74% | 0 | 0 | — |
case-13 | fail→pass | 20,696 | 17,292 | -16% | 1 | 1 | 0% | 3,185 | 3,213 | +1% | 0 | 0 | — |
case-14 | pass→pass | 19,068 | 28,282 | +48% | 1 | 1 | 0% | 2,988 | 5,197 | +74% | 0 | 0 | — |
case-15 | pass→pass | 22,074 | 30,320 | +37% | 1 | 1 | 0% | 3,581 | 5,685 | +59% | 0 | 0 | — |
case-16 | pass→pass | 22,273 | 36,678 | +65% | 1 | 1 | 0% | 4,074 | 6,428 | +58% | 0 | 0 | — |
case-17 | pass→pass | 24,081 | 33,420 | +39% | 1 | 1 | 0% | 4,078 | 5,781 | +42% | 0 | 0 | — |
case-18 | pass→pass | 20,978 | 21,571 | +3% | 1 | 1 | 0% | 3,370 | 4,423 | +31% | 0 | 0 | — |
case-19 | pass→pass | 18,634 | 24,079 | +29% | 1 | 1 | 0% | 2,945 | 4,586 | +56% | 0 | 0 | — |
case-20 | pass→fail | 9,710 | 13,763 | +42% | 1 | 1 | 0% | 1,468 | 2,586 | +76% | 0 | 0 | — |
case-22 | pass→pass | 15,919 | 15,756 | -1% | 1 | 1 | 0% | 3,072 | 3,560 | +16% | 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. 2 cases got worse with the skill loaded, and they are 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.