Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Interactive StyleSeed setup wizard for choosing app type, brand color, visual style, typography, and the first screen scaffold.
.claude/skills/lingxling-ui-setup/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -75% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -79% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -83% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -35% | 0% |
Part of StyleSeed, this setup wizard turns a raw project into a design-system-guided workspace. It collects the minimum brand and product context needed to configure tokens, pick a visual direction, and generate an initial page without drifting into generic UI.
Do not front-load the full questionnaire. Ask a single question, wait for the answer, store it, then continue.
Identify the product shape before touching tokens or layout recipes.
Suggested buckets:
Use the answer to choose the page composition pattern and the type of first screen to scaffold.
Offer a few safe defaults plus a custom hex option. Once selected:
If the project uses the StyleSeed Toss seed, the main target is css/theme.css.
Ask whether the user wants to borrow the feel of an established brand or design language. Good examples include Stripe, Linear, Vercel, Notion, Spotify, Supabase, and Airbnb.
Use the reference to influence density, tone, and composition, not to clone assets or trademarks.
Confirm the font direction:
If the seed is present, update the font-related files rather than scattering overrides across components.
Ask for:
Then scaffold the page using the seed's page shell, top bar, navigation, spacing scale, and card structure.
Return:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 25,763 | 13,867 | -46% | 1 | 1 | 0% | 5,383 | 1,354 | -75% | 0 | 0 | — |
case-02 | fail→pass | 41,406 | 4,548 | -89% | 1 | 1 | 0% | 7,742 | 1,622 | -79% | 0 | 0 | — |
case-03 | fail→pass | 44,521 | 4,634 | -90% | 1 | 1 | 0% | 8,261 | 1,408 | -83% | 0 | 0 | — |
case-04 | pass→pass | 13,715 | 11,590 | -15% | 1 | 1 | 0% | 2,094 | 2,453 | +17% | 0 | 0 | — |
case-05 | pass→pass | 11,757 | 6,689 | -43% | 1 | 1 | 0% | 1,756 | 1,986 | +13% | 0 | 0 | — |
case-06 | pass→pass | 14,827 | 11,018 | -26% | 1 | 1 | 0% | 2,457 | 3,058 | +24% | 0 | 0 | — |
case-07 | fail→fail | 13,973 | 15,444 | +11% | 1 | 1 | 0% | 1,972 | 2,625 | +33% | 0 | 0 | — |
case-08 | fail→pass | 15,835 | 2,976 | -81% | 1 | 1 | 0% | 2,663 | 1,081 | -59% | 0 | 0 | — |
case-09 | pass→pass | 16,841 | 10,322 | -39% | 1 | 1 | 0% | 2,385 | 2,290 | -4% | 0 | 0 | — |
case-10 | pass→pass | 12,695 | 7,701 | -39% | 1 | 1 | 0% | 1,919 | 1,773 | -8% | 0 | 0 | — |
case-11 | pass→pass | 13,420 | 12,191 | -9% | 1 | 1 | 0% | 2,490 | 2,977 | +20% | 0 | 0 | — |
case-12 | fail→pass | 14,589 | 4,167 | -71% | 1 | 1 | 0% | 1,875 | 1,226 | -35% | 0 | 0 | — |
case-13 | fail→pass | 10,783 | 5,764 | -47% | 1 | 1 | 0% | 1,660 | 1,524 | -8% | 0 | 0 | — |
case-14 | fail→fail | 7,966 | 3,430 | -57% | 1 | 1 | 0% | 1,209 | 1,324 | +10% | 0 | 0 | — |
case-15 | pass→pass | 17,024 | 7,085 | -58% | 1 | 1 | 0% | 2,765 | 1,879 | -32% | 0 | 0 | — |
case-16 | fail→pass | 10,442 | 6,921 | -34% | 1 | 1 | 0% | 1,639 | 1,796 | +10% | 0 | 0 | — |
case-17 | fail→pass | 13,678 | 8,601 | -37% | 1 | 1 | 0% | 1,998 | 1,949 | -2% | 0 | 0 | — |
case-18 | pass→pass | 14,109 | 6,477 | -54% | 1 | 1 | 0% | 1,909 | 1,592 | -17% | 0 | 0 | — |
case-19 | pass→pass | 15,449 | 15,062 | -3% | 1 | 1 | 0% | 2,433 | 2,772 | +14% | 0 | 0 | — |
case-20 | fail→fail | 7,442 | 4,275 | -43% | 1 | 1 | 0% | 1,208 | 1,373 | +14% | 0 | 0 | — |
case-21 | pass→pass | 15,445 | 3,137 | -80% | 1 | 1 | 0% | 2,147 | 1,237 | -42% | 0 | 0 | — |
case-22 | pass→pass | 14,553 | 7,480 | -49% | 1 | 1 | 0% | 1,943 | 2,187 | +13% | 0 | 0 | — |
case-23 | fail→fail | 22,809 | 19,194 | -16% | 1 | 1 | 0% | 5,448 | 4,919 | -10% | 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. The headline lift of +35 percentage points is the difference between those two pass rates over the 23 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.