Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A pixel-faithful, self-contained mirror of the live open-design.ai homepage — an interactive React Three Fiber / Next.js hero with a real-time 3D wordmark, sticker collage, variable fonts, and scroll-driven motion. First-party showcase of the visual ceiling for interactive web marketing surfaces.
.claude/skills/nexu-io-open-design-homepage/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -29% | 0% |
A first-party showcase template that mirrors the live open-design.ai homepage, captured as a fully self-contained bundle that renders in the sandboxed preview.
It is here to demonstrate the visual ceiling OpenDesign targets for interactive, WebGL-grade marketing surfaces on the web — not a fill-in-the-blank generator. Treat it as a reference build to study and adapt.
OpenDesign script wordmark rendered with React Three Fiber / Three.js, plus a cursor-reactive parallax layer and a floating sticker collage.KHR_draco_mesh_compression) — the hero wordmark drops from ~4.4 MB to ~300 KB with no visible change, decoded at runtime by the bundled DRACOLoader.Open example.html in the preview to explore the live, interactive result. All content is bundled locally — fonts, imagery, and 3D geometry ship in assets/, so there are no CDN scripts or external fonts. The one runtime network request is the Draco decoder, fetched from Google's gstatic CDN on demand (the same pattern the other WebGL examples use for their libraries). To adapt the aesthetic (typography, 3D treatment, sticker collage, scroll motion) into your own build, study the structure here and recreate it with your preferred React Three Fiber / Next.js setup.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 17,591 | 23,476 | +33% | 1 | 1 | 0% | 2,710 | 2,549 | -6% | 0 | 0 | — |
case-02 | fail→fail | 17,679 | 25,813 | +46% | 1 | 1 | 0% | 2,773 | 4,343 | +57% | 0 | 0 | — |
case-03 | fail→pass | 19,193 | 15,832 | -18% | 1 | 1 | 0% | 3,219 | 3,264 | +1% | 0 | 0 | — |
case-04 | fail→pass | 9,881 | 4,443 | -55% | 1 | 1 | 0% | 1,705 | 1,271 | -25% | 0 | 0 | — |
case-18 | fail→pass | 10,952 | 9,155 | -16% | 1 | 1 | 0% | 1,766 | 1,743 | -1% | 0 | 0 | — |
case-05 | fail→pass | 13,109 | 4,408 | -66% | 1 | 1 | 0% | 1,681 | 1,191 | -29% | 0 | 0 | — |
case-06 | fail→pass | 19,922 | 2,511 | -87% | 1 | 1 | 0% | 2,800 | 909 | -68% | 0 | 0 | — |
case-07 | fail→pass | 14,253 | 4,590 | -68% | 1 | 1 | 0% | 1,859 | 1,323 | -29% | 0 | 0 | — |
case-08 | fail→pass | 11,187 | 3,001 | -73% | 1 | 1 | 0% | 1,990 | 916 | -54% | 0 | 0 | — |
case-19 | pass→pass | 13,751 | 3,310 | -76% | 1 | 1 | 0% | 1,812 | 1,052 | -42% | 0 | 0 | — |
case-09 | fail→pass | 12,370 | 3,575 | -71% | 1 | 1 | 0% | 1,371 | 1,070 | -22% | 0 | 0 | — |
case-10 | fail→pass | 14,422 | 2,652 | -82% | 1 | 1 | 0% | 2,277 | 965 | -58% | 0 | 0 | — |
case-11 | fail→pass | 7,499 | 2,107 | -72% | 1 | 1 | 0% | 1,017 | 764 | -25% | 0 | 0 | — |
case-12 | pass→pass | 12,583 | 3,650 | -71% | 1 | 1 | 0% | 1,707 | 1,009 | -41% | 0 | 0 | — |
case-13 | fail→pass | 12,386 | 2,322 | -81% | 1 | 1 | 0% | 1,594 | 843 | -47% | 0 | 0 | — |
case-14 | fail→fail | 10,054 | 2,979 | -70% | 1 | 1 | 0% | 1,404 | 834 | -41% | 0 | 0 | — |
case-15 | fail→fail | 10,958 | 7,981 | -27% | 1 | 1 | 0% | 2,047 | 1,663 | -19% | 0 | 0 | — |
case-16 | fail→pass | 9,001 | 2,858 | -68% | 1 | 1 | 0% | 1,177 | 983 | -16% | 0 | 0 | — |
case-17 | pass→pass | 7,061 | 3,783 | -46% | 1 | 1 | 0% | 1,231 | 968 | -21% | 0 | 0 | — |
case-20 | pass→pass | 11,906 | 8,540 | -28% | 1 | 1 | 0% | 1,784 | 2,086 | +17% | 0 | 0 | — |
case-21 | pass→pass | 9,929 | 15,253 | +54% | 1 | 1 | 0% | 1,526 | 2,446 | +60% | 0 | 0 | — |
case-22 | pass→pass | 11,420 | 9,974 | -13% | 1 | 1 | 0% | 2,045 | 2,014 | -2% | 0 | 0 | — |
case-23 | fail→fail | 11,464 | 2,853 | -75% | 1 | 1 | 0% | 1,592 | 1,001 | -37% | 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 +57 percentage points is the difference between those two pass rates over the 23 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.