Skip to content

Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.

4.1 KB
Raw Blame

@three-ws/create-agent

One command from a sentence to a rigged 3D character you can put on your site.

npm create @three-ws/agent "a friendly cartoon astronaut in a glossy white suit"
three.ws building "a friendly cartoon astronaut in a glossy white suit"
  a rigged humanoid, free lane, no account needed

· sent to the three.ws forge
· generating the figure (15s)
· generating the figure, then the skeleton (60s)
✓ model ready
· downloading the model

Friendly Cartoon Astronaut is ready in 94s

  folder  a-friendly-cartoon-astronaut/
          index.html
          agent.json
          README.md
          agent.glb (3.4 MB)
  viewer  https://three.ws/viewer?src=…

  Paste this anywhere:

  <script type="module" src="https://three.ws/agent-3d/latest/agent-3d.js"></script>

  <agent-3d body="https://three.ws/cdn/…/rigged.glb"
            name="Friendly Cartoon Astronaut"
            style="width:100%;max-width:420px;height:520px"></agent-3d>

  Open the demo:  npx serve a-friendly-cartoon-astronaut

No account. No API key. No signup. The generation runs on the three.ws free lane, and the platform's own provider keys cover it.

What you get

File What it is
agent.glb The model, with a humanoid skeleton, so it can be posed and animated. Opens in Blender, Unity, Godot, or any glTF tool.
index.html A working demo page. Double click it.
agent.json The record: prompt, hosted URLs, backend, timing.
README.md How to embed it and where to go next.

Options

npm create @three-ws/agent "a knight in worn steel armor" --out ./knight
npm create @three-ws/agent --photo https://example.com/me.jpg --name "Me"
npm create @three-ws/agent "a small ceramic frog figurine" --object
Option Meaning
--photo <url> Build from a public https reference image instead of a description
--out <dir> Where to write the project (default: a slug of the description)
--name <name> Display name (default: derived from the description)
--object An object or prop (mesh only, free lane) instead of a rigged character
--no-download Keep the hosted model URL, skip writing agent.glb
--json Print the result as JSON and nothing else
--origin <url> API origin (default https://three.ws)

Describe one full-body humanoid in a neutral standing pose for the best rig. Lead with the subject, then materials and colours: the free lane conditions on roughly the first 77 characters, so front-load what matters.

Programmatic use

import { createAgent } from '@three-ws/create-agent';

const made = await createAgent({
	prompt: 'a friendly cartoon astronaut',
	out: './astro',
	onProgress: ({ message }) => console.log(message),
});

made.result.glbUrl; // hosted, durable
made.result.rigged; // true
made.dir; // absolute path to the project
Export What it does
createAgent(opts) The whole pipeline: generate, scaffold, download
callForge({ tool, args, ... }) One forge call (forge_avatar or forge_free)
downloadModel(url) Fetch the model bytes with a deadline
projectFiles(...), demoPage(...), embedSnippet(...) The scaffold, as pure functions
ForgeError Carries code: timeout, unreachable, no_model, rpc_error, http_<status>

Everything takes { fetchImpl, signal, origin }, so it runs against a local three.ws, a proxy, or a test double.

How long it takes

A rigged character is generation plus rigging, usually one to three minutes; rigging is the slow half. An --object is faster. Every call carries an explicit deadline (10 minutes) and reports real elapsed time, never an invented progress bar. If the deadline hits, the job may still finish: finished models land at https://three.ws/creations.

Related

License

Apache-2.0