Agent skills that teach Claude Code, Codex and other AI coding agents to build 3D product viewers, configurators, AR embeds and 360° spins, with tested templates, current three.js APIs and real performance budgets.
All three demos show the same product. The model is assets/mug-optimized.glb (31 KB after glTF-Transform), exported from three.js by tools/export_mug.py.
| Skill | What your agent learns |
|---|---|
threejs-product-viewer | Load a GLB, auto-frame any model, studio lighting, contact shadow, orbit limits, render on demand, cleanup and budgets |
product-configurator | Colors, finishes, textures, parts and KHR_materials_variants; URL state, pricing, accessible option UI |
3d-asset-optimization | glTF-Transform: Meshopt vs Draco, KTX2 vs WebP, budgets, and the flags that silently break configurators |
model-viewer-ar | <model-viewer> embeds with AR on Android and iOS, true-scale checks, variants, custom AR button |
360-product-spin | When to use a spin, how to photograph or render the frames, and embedding with Spin360 |
/plugin marketplace add ahmettasdemirusa/threejs-product-skills /plugin install threejs-product-skills@threejs-product-skills
In Claude Code. For other agents, copy the skills/ folders; they use the open Agent Skills (SKILL.md) format.