# Add to your Claude Code skills
git clone https://github.com/codepawl/67.jsGuides for using ai agents skills like 67.js.
Last scanned: 11/14/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-11-14T11:15:13.229Z",
"npmAuditRan": true,
"pipAuditRan": true,
"promptInjectionRan": true
}See how 67.js compares with popular alternatives.
67.js is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by codepawl. 3D engine that AI agents drive. It has 624 GitHub stars.
Yes. 67.js passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/codepawl/67.js" and add it to your Claude Code skills directory (see the Installation section above).
67.js is primarily written in JavaScript. It is open-source under codepawl on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh 67.js against similar tools.
No comments yet. Be the first to share your thoughts!
From $49 · 30 days · terms & refunds
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
Automated security scan — not a certification.
67js is a 3D engine for JavaScript that an AI agent drives through an MCP server. It makes 3D easy to understand and easy to do.
The agent knows every part of 67js and carries the heavy steps. It builds the scene, lights it, checks the preview and renders the result. You keep the creative choices and stay in the loop, because the agent shows previews and asks when a matter of taste is open.
More scenes are on the showcase page and in the Showcase section below.
67js is a 3D engine for AI agents and for pages. It has a real-time rasterizer, a path tracer, addons for terrain, vegetation, characters, liquid, vehicles and media export, and an MCP server. A scene is plain data: each part has a geometry, a matrix and a material.
An agent builds and changes a scene through the MCP server, looks at PNG previews and renders stills or MP4 video. The goal is to be as capable as Blender and as light as three.js. docs/goal.md defines each part. See the documentation, the guide for agents and the FAQ.
Let your agent do it. Paste this prompt into Claude Code, Codex, Cursor, Gemini CLI or OpenCode. The agent installs the package, reads the guide and connects the MCP server.
Install 67js from npm as `67js@npm:@codepawlhq/67js` and connect its MCP server with `npx 67js-mcp`. Read
https://67js.codepawl.com/llms.txt first and follow it.
Claude Code plugin. One command adds the MCP server and both skills.
/plugin marketplace add codepawl/67.js
/plugin install 67js@67js
By hand
npm install 67js@npm:@codepawlhq/67js
The registry name is @codepawlhq/67js, and the alias 67js keeps imports short. The package is ESM and needs Node.js 20.19 or later. Glaze and the WebGL2 path tracer need a WebGL2 browser, and the WebGPU backends need WebGPU.
A new minor version ships about once a week. Before 1.0, a minor version can change the API, and CHANGELOG.md lists each change. npx keeps the copy it downloaded first. To update, run npm install 67js@npm:@codepawlhq/67js@latest or npx -y -p @codepawlhq/67js@latest 67js-mcp.
import { Renderer67, box, sphere } from '67js';
// Column-major 4x4 matrix: scale, then move.
const placed = ([x, y, z], [width, height, depth] = [1, 1, 1]) =>
new Float32Array([width, 0, 0, 0, 0, height, 0, 0, 0, 0, depth, 0, x, y, z, 1]);
const parts = [
{
name: 'Floor',
geometry: box(),
matrix: placed([0, 0, 0], [8, 0.1, 8]),
material: { color: [0.4, 0.42, 0.45], metallic: 0, roughness: 0.9, emission: 0 },
},
{
name: 'Ball',
geometry: sphere(48, 24),
matrix: placed([0, 1, 0]),
material: { color: [0.8, 0.3, 0.2], metallic: 0, roughness: 0.35, emission: 0 },
},
];
const canvas = document.querySelector('canvas');
const renderer = new Renderer67(canvas, parts, { samples: 4, memoryBudgetMiB: 256 });
renderer.render({ eye: [4, 3, 6], target: [0, 1, 0] });
Call renderer.updateParts(parts) to move or recolour parts without a new upload.
Call renderer.resize() after the canvas changes size, and renderer.dispose() before you replace the renderer.
The guide covers cameras, lights, effects and every material field.
To save a PNG from a script, use the render-from-a-script recipe.
The package includes an MCP server that connects an agent to a persistent browser renderer. The server needs a local Chrome installation.
npx 67js-mcp --backend gpu --memory 192
--backend gpu renders with hardware WebGL2 through Chrome. Use --backend cpu on a machine without a GPU.
With bun, bunx 67js-mcp takes the same flags; CI runs the packaged server under bun on every pull request.
An agent authors a JSON recipe, sets lights and fog, imports GLB assets and adds motion.
It gets PNG previews back, compares up to four views in a review grid and runs render jobs for stills and MP4 video.
The server accepts no arbitrary code and no remote asset URLs. The MCP reference lists its tools.
Two skills teach an agent to drive the MCP tools: build-67js-scene for scenes and materials, and render-67js-media for image and video export. For a plain-text guide, give the agent llms.txt.
viz3d uses 67js. It is a skill that turns places, systems and events into 3D apps and films, and it renders its films and stills in 67js.
To add it in Claude Code, run /plugin marketplace add codepawl/viz3d, then /plugin install visualize-in-3d@codepawl.
Each picture below is a 67js render. The linked ones open live in the browser.
The render command exports a scene to PNG or MP4 through headless Chrome, on the GPU or on the CPU.
npm run render -- --kind image --scene nature --backend gpu --preset high --output artifacts/nature.png
Video export streams one frame at a time, so a film never sits in memory. The render skill lists every preset and flag.
Each addon is a separate import, so a page loads only what it uses.
| Area | Import | Guide |
|---|---|---|
| Lighting | 67js/addons/hdri |
Image-based lighting from .hdr files |
| Terrain | 67js/addons/terrain, 67js/addons/texture-set |
Chunked terrain with scanned ground layers |
| Vegetation | 67js/addons/vegetation, 67js/addons/grass, 67js/addons/clutter |
[Trees](ht |