Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A scroll-reactive masonry gallery (Three.js + GSAP) where each image develops from a pixel/grid dissolve on viewport entry, with a split-text heading and click-to-fullscreen Flip.
.claude/skills/nexu-io-webgl-pixel-reveal-gallery/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✓→✗ | ▼ Worse | 97% | 0% |
| case-07 | ✓→✗ | ▼ Worse | 3% | 0% |
| case-09 | ✓→✗ | ▼ Worse | 3% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 268% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 78% | 0% |
Produce a single self-contained index.html — A scroll-reactive masonry gallery (Three.js + GSAP) where each image develops from a pixel/grid dissolve on viewport entry, with a split-text heading and click-to-fullscreen Flip.
OpenDesign detects getContext('webgl2') / heavy WebGL and renders this file in powered preview (a cross-origin-isolated iframe). The full GPU + scroll pipeline runs; no opaque-sandbox workarounds are needed.
webgl-pixel-reveal-gallery/
├── SKILL.md ← you're reading this
├── example.html ← the complete, working artifact (READ FIRST)
└── (assets, if any)Keep any bundled LICENSE and on-screen credit intact. Replace imagery only with license-clean assets (original / AI, Lummi.ai, Unsplash/Pexels — never scraped imagery).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | pass→fail | 21,470 | 47,429 | +121% | 1 | 1 | 0% | 4,298 | 8,466 | +97% | 0 | 0 | — |
case-12 | fail→fail | 38,917 | 51,777 | +33% | 1 | 1 | 0% | 7,834 | 8,463 | +8% | 0 | 0 | — |
case-01 | fail→fail | 44,073 | 45,811 | +4% | 1 | 1 | 0% | 8,259 | 8,504 | +3% | 0 | 0 | — |
case-02 | fail→fail | 43,309 | 41,855 | -3% | 1 | 1 | 0% | 8,251 | 8,495 | +3% | 0 | 0 | — |
case-03 | fail→fail | 41,847 | 42,746 | +2% | 1 | 1 | 0% | 8,248 | 8,492 | +3% | 0 | 0 | — |
case-04 | pass→pass | 11,524 | 44,266 | +284% | 1 | 1 | 0% | 2,300 | 8,472 | +268% | 0 | 0 | — |
case-06 | pass→pass | 21,696 | 37,783 | +74% | 1 | 1 | 0% | 4,747 | 8,464 | +78% | 0 | 0 | — |
case-07 | pass→fail | 38,686 | 43,230 | +12% | 1 | 1 | 0% | 8,217 | 8,461 | +3% | 0 | 0 | — |
case-08 | fail→fail | 45,995 | 46,022 | +0% | 1 | 1 | 0% | 8,221 | 8,465 | +3% | 0 | 0 | — |
case-09 | pass→fail | 54,716 | 44,117 | -19% | 1 | 1 | 0% | 8,224 | 8,468 | +3% | 0 | 0 | — |
case-10 | fail→fail | 43,653 | 41,876 | -4% | 1 | 1 | 0% | 8,231 | 8,475 | +3% | 0 | 0 | — |
case-11 | fail→fail | 43,042 | 44,587 | +4% | 1 | 1 | 0% | 8,219 | 8,463 | +3% | 0 | 0 | — |
case-13 | fail→fail | 43,761 | 42,721 | -2% | 1 | 1 | 0% | 8,220 | 8,464 | +3% | 0 | 0 | — |
case-14 | fail→fail | 42,560 | 43,626 | +3% | 1 | 1 | 0% | 8,220 | 8,464 | +3% | 0 | 0 | — |
case-15 | fail→fail | 42,608 | 40,996 | -4% | 1 | 1 | 0% | 8,218 | 8,462 | +3% | 0 | 0 | — |
case-16 | fail→fail | 24,937 | 38,355 | +54% | 1 | 1 | 0% | 5,040 | 8,463 | +68% | 0 | 0 | — |
case-17 | fail→fail | 42,662 | 39,363 | -8% | 1 | 1 | 0% | 8,218 | 8,462 | +3% | 0 | 0 | — |
case-18 | fail→fail | 40,159 | 44,100 | +10% | 1 | 1 | 0% | 8,210 | 8,454 | +3% | 0 | 0 | — |
case-19 | fail→fail | 43,797 | 44,177 | +1% | 1 | 1 | 0% | 8,225 | 8,469 | +3% | 0 | 0 | — |
case-20 | fail→fail | 43,238 | 43,159 | -0% | 1 | 1 | 0% | 8,220 | 8,464 | +3% | 0 | 0 | — |
case-21 | fail→fail | 40,972 | 41,640 | +2% | 1 | 1 | 0% | 8,216 | 8,460 | +3% | 0 | 0 | — |
case-22 | fail→fail | 10,134 | 4,766 | -53% | 1 | 1 | 0% | 1,600 | 965 | -40% | 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 -14 percentage points is the difference between those two pass rates over the 22 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.