Releases: Impertio-Studio/Three.js-Claude-Skill-Package
Release list
v1.1.1: Improved Skill Discoverability
What's new
Expanded Keywords in all skill descriptions with user-friendly, non-technical discovery terms. Skills are now easier to find when users describe problems in plain language instead of technical terminology.
Examples of added terms:
- Symptom-based: "app is slow", "nothing shows", "container won't start"
- Beginner-friendly: "how do I start", "what is", "getting started"
- Error messages: literal error text users would paste
- Alternative names: common synonyms and abbreviations
No skill content was changed. Only the Keywords line in SKILL.md frontmatter was expanded.
v1.1.0: Discovery Manifests
What's new
- package.json with agents.skills[] manifest (24 skills) for external discovery
- agents/openai.yaml for OpenAI Codex
- GitHub topic
agentskillsadded - SKILLS_LOG.md added to .gitignore
- Em-dashes replaced with colons in 4 section headings
Enables discovery via npx skills add, skills.sh, LobeHub, and other platforms.
v1.0.0 — 24 Deterministic Three.js Skills
Three.js Claude Skill Package v1.0.0
24 deterministic skills for Three.js 3D web development with Claude Code.
Skills (24 total)
| Category | Count | Skills |
|---|---|---|
| Core | 4 | scene-graph, renderer, math, raycaster |
| Syntax | 5 | geometries, materials, shaders, loaders, controls |
| Implementation | 10 | lighting, shadows, animation, post-processing, physics, react-three-fiber, drei, webgpu, ifc-viewer, audio, xr |
| Errors | 2 | performance, rendering |
| Agents | 2 | scene-builder, model-optimizer |
Quality
- 24/24 skills pass all validation checks
- Average 338 lines per SKILL.md (all under 500)
- 96 reference files (methods, examples, anti-patterns)
- 27,627 words of research documentation
- 220+ anti-patterns documented
- Zero hedging words — deterministic ALWAYS/NEVER language
- English-only, ES module imports, Three.js r160+
Technology Coverage
Three.js r160+, React Three Fiber 8.x, Drei, WebGPU/TSL, cannon-es, Rapier, WebXR VR/AR, IFC/BIM, 3D Audio
Methodology
Built with the 7-phase research-first methodology in a single session: landscape mapping → 27,627 words research → 24 atomic skills → 8 parallel batches → validation → publication.