cd /news/ai-agents/open-source-67-js-gives-ai-agents-a-… · home › topics › ai-agents › article
[ARTICLE · art-169368] src=alphasignal.ai ↗ pub= topic=ai-agents verified=true sentiment=↑ positive

Open-Source 67.js Gives AI Agents a Full 3D Rendering Toolbox

Codepawl released 67.js, an MIT-licensed WebGL2 3D engine whose bundled Model Context Protocol server exposes 25 tools so AI agents can create scenes, import GLBs, and render PNG stills or MP4 video. Scenes are plain JSON data with geometry, transformation matrices, and materials per part, and the engine ships two backends: the Glaze rasterizer for real-time work and the Cook path tracer for finished stills. A lit-box demo is 58.8 kB gzipped versus 133.8 kB for three.js, 2.28x smaller; the package installs via `npm i @codepawlhq/67js` and requires Node.js 20.19 and WebGL2.

read2 min views1 publishedNov 13, 2026
Open-Source 67.js Gives AI Agents a Full 3D Rendering Toolbox
Image: source

67.js is an MIT-licensed WebGL2 3D engine with an MCP server exposing 25 tools, letting AI agents build scenes, import GLBs and render MP4 video.

  • 67.js is an MIT-licensed WebGL2 3D engine built for AI agents to drive.
  • Ships an MCP server exposing 25 tools for scene editing, asset import and rendering.
  • Two backends: Glaze rasterizer for real-time, Cook path tracer for finished stills.
  • Scenes are plain JSON data: geometry, matrix and material per part.
  • Lit-box demo is 58.8 kB gzipped versus 133.8 kB for three.js, 2.28x smaller.
  • Install via npm i @codepawlhq/67js , requires Node.js 20.19 and WebGL2.

67.js is an MIT-licensed WebGL2 engine built around AI agents as its primary users. Its bundled Model Context Protocol server exposes scene creation, asset import, lighting, animation and rendering as callable tools. Any MCP client that supports standard input and output can use those tools to produce PNG stills or MP4 video.

This approach gives automated workflows a direct 3D authoring interface. An agent sends declarative scene data to a persistent headless Chrome process, reviews rendered previews and revises the scene without operating a graphical modeling application.

Runtime and distribution
Requirement Details
--- ---
Package @codepawlhq/67js on npm
License MIT
Runtime Node.js 20.19 or later
Rendering A WebGL2-capable browser or headless Chrome
MCP transport Local process over stdio
Service model Local installation with no hosted API or pricing tier

After installing the package, developers launch the bundled MCP server and register that local process in an MCP client's configuration. The renderer remains on the developer's machine, along with the scene files and imported assets.

JSON becomes the scene graph #

Each 67.js scene is plain data composed of parts with geometry, transformation matrices and materials. Agents generate and revise JSON recipes that describe those parts alongside lights, cameras and environmental settings. The engine also provides modules for terrain, vegetation, characters, liquids, vehicles and media export.

This story is for Pro members #

You've reached the end of the free preview. Upgrade to AlphaSignal Pro to read the full article - and everything else behind the paywall.

── more in #ai-agents 4 stories · sorted by recency
── more on @67.js 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/open-source-67-js-gi…] indexed:0 read:2min 2026-11-13 · —