Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A real-time raymarched 3D hero — soft-shadowed morphing metaballs with a fresnel rim light, sphere-traced per pixel in a fragment shader with a slow camera orbit. No meshes, no textures. Rendered as a single self-contained `index.html`. Use when the brief asks for a "raymarch", "signed distance field", "SDF", "3D hero", "metaballs", "blobs", or a shader-based 3D scene. Open Design serves this in powered-preview mode so the GPU stack actually runs.
.claude/skills/nexu-io-webgl-raymarched-hero/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 39% | 0% |
| case-07 | ✓→✗ | ▼ Worse | 184% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 8% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 8% | 0% |
Produce a single self-contained index.html that renders a real-time 3D scene entirely in a fragment shader via sphere tracing, full-screen, with a clean typographic overlay on top.
OpenDesign detects getContext('webgl2') and renders this file in powered preview — a cross-origin-isolated iframe with allow-same-origin. The full GPU pipeline is available; you do not need to work around the opaque sandbox.
webgl-raymarched-hero/
├── SKILL.md ← you're reading this
└── example.html ← a working SDF metaball raymarcher (READ FIRST)Read example.html end to end. Note the shape: one fullscreen triangle; a map(p) signed-distance function that smooth-unions (smin) several moving spheres; normals by finite differences; a softShadow march; and a camera built from ro/ta with a slow orbit. Shading = diffuse × soft shadow + fresnel rim + a tight specular.
Keep the marcher cheap so it holds 60fps:
smin — organic, always readable.index.htmld < 0.001) and on far distance; clamp DPR to ~1.5 — raymarching is per-pixel heavy.pow(1 - dot(n,-rd), 3)), and one specular lobe. Gentle gamma at the end.#63fe13 and rims it in teal).webgl2 is unavailable.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 52,037 | 58,800 | +13% | 1 | 1 | 0% | 8,274 | 8,953 | +8% | 0 | 0 | — |
case-02 | fail→fail | 41,372 | 52,592 | +27% | 1 | 1 | 0% | 8,267 | 8,946 | +8% | 0 | 0 | — |
case-03 | fail→fail | 40,090 | 50,829 | +27% | 1 | 1 | 0% | 8,257 | 8,936 | +8% | 0 | 0 | — |
case-04 | pass→pass | 37,173 | 57,687 | +55% | 1 | 1 | 0% | 8,232 | 8,911 | +8% | 0 | 0 | — |
case-05 | pass→pass | 87,201 | 52,147 | -40% | 1 | 1 | 0% | 8,233 | 8,912 | +8% | 0 | 0 | — |
case-06 | pass→fail | 40,085 | 41,011 | +2% | 1 | 1 | 0% | 6,407 | 8,894 | +39% | 0 | 0 | — |
case-07 | pass→fail | 17,931 | 42,435 | +137% | 1 | 1 | 0% | 3,146 | 8,932 | +184% | 0 | 0 | — |
case-08 | fail→pass | 15,804 | 7,314 | -54% | 1 | 1 | 0% | 2,866 | 1,895 | -34% | 0 | 0 | — |
case-09 | pass→pass | 17,358 | 16,990 | -2% | 1 | 1 | 0% | 3,038 | 3,285 | +8% | 0 | 0 | — |
case-10 | fail→fail | 19,949 | 15,367 | -23% | 1 | 1 | 0% | 2,836 | 3,318 | +17% | 0 | 0 | — |
case-11 | pass→pass | 25,565 | 50,925 | +99% | 1 | 1 | 0% | 4,683 | 8,901 | +90% | 0 | 0 | — |
case-12 | pass→pass | 29,641 | 40,323 | +36% | 1 | 1 | 0% | 3,204 | 8,903 | +178% | 0 | 0 | — |
case-13 | fail→fail | 16,889 | 16,404 | -3% | 1 | 1 | 0% | 2,521 | 3,234 | +28% | 0 | 0 | — |
case-14 | pass→pass | 37,035 | 39,921 | +8% | 1 | 1 | 0% | 8,236 | 8,915 | +8% | 0 | 0 | — |
case-15 | pass→pass | 11,021 | 10,065 | -9% | 1 | 1 | 0% | 2,177 | 2,809 | +29% | 0 | 0 | — |
case-16 | pass→pass | 18,657 | 11,990 | -36% | 1 | 1 | 0% | 2,850 | 3,151 | +11% | 0 | 0 | — |
case-17 | pass→pass | 14,814 | 15,923 | +7% | 1 | 1 | 0% | 2,560 | 3,580 | +40% | 0 | 0 | — |
case-18 | pass→pass | 23,865 | 10,681 | -55% | 1 | 1 | 0% | 3,854 | 2,807 | -27% | 0 | 0 | — |
case-19 | pass→pass | 19,316 | 12,478 | -35% | 1 | 1 | 0% | 3,063 | 2,865 | -6% | 0 | 0 | — |
case-20 | pass→pass | 31,188 | 55,430 | +78% | 1 | 1 | 0% | 6,588 | 8,909 | +35% | 0 | 0 | — |
case-21 | pass→pass | 12,768 | 13,722 | +7% | 1 | 1 | 0% | 2,638 | 3,223 | +22% | 0 | 0 | — |
case-22 | pass→pass | 22,667 | 17,873 | -21% | 1 | 1 | 0% | 3,318 | 3,246 | -2% | 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 -5 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.