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.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
User-agent: *
Allow: /
Disallow: /api/
# The PNG render of a 3D model. Creation pages name it as their og:image,
# twitter:image and JSON-LD thumbnailUrl, so every reader that falls to this
# group (X's card crawler, Grok, image search) has to be able to fetch it.
Allow: /api/render/glb
Disallow: /studio-deposit-harness*
# /dashboard, /settings and /my-agents are signed-in surfaces and every one of
# their pages carries <meta name="robots" content="noindex">. A Disallow here
# would hide that tag from the crawler that needs to read it: Google kept them
# in the index as bare URLs ("Indexed, though blocked by robots.txt") precisely
# because it was never allowed to fetch the page and see the noindex. Crawling
# is what removes them, so leave these open.
# Embed routes serve raw runtime, not human content; keep them out of SERPs.
Disallow: /agent/*/embed
Disallow: /widget/*/embed
Disallow: /avatar/*/embed
# Cosmetic OG image endpoints: fine for social crawlers, useless for SERP.
Disallow: /api/agent/*/og
Disallow: /api/avatar/*/og
# ----------------------------------------------------------------------
# AI assistants fetching on behalf of a live user (ChatGPT, Claude,
# Perplexity, DuckDuckGo, Mistral) are welcome everywhere public,
# INCLUDING the API: the x402 pay-per-call endpoints, the free 3D lanes,
# and the MCP servers are the product. Only signed-in account surfaces
# stay off limits.
#
# Grok is welcome too, but xAI documents no user-agent token for its
# crawler or its live-user fetcher (checked against docs.x.ai on
# 2026-10-08), and Grok Bot browses with an ordinary browser. Listing an
# undocumented token here would match nothing real, so Grok falls to the
# "User-agent: *" group above, which serves it every public page and the
# model renders. When xAI publishes its tokens, add them to this group.
# ----------------------------------------------------------------------
User-agent: ChatGPT-User
User-agent: OAI-SearchBot
User-agent: Claude-User
User-agent: Claude-SearchBot
User-agent: Perplexity-User
User-agent: DuckAssistBot
User-agent: MistralAI-User
Allow: /
Disallow: /my-agents
Disallow: /dashboard
Disallow: /settings
Disallow: /studio-deposit-harness*
# AI index and training crawlers (GPTBot, ClaudeBot, PerplexityBot,
# Google-Extended, Bytespider, CCBot, xAI's, and the rest) are equally welcome and
# already covered by the "User-agent: *" rules above; nothing here blocks
# them. Start from the machine-readable index instead of scraping HTML:
#
# https://three.ws/llms.txt concise AI index (https://llmstxt.org/)
# https://three.ws/llms-full.txt long-form page index
# https://three.ws/.well-known/mcp.json hosted MCP servers
# https://three.ws/.well-known/agent-card.json A2A agent card
# https://three.ws/.well-known/x402.json paid-service catalog (x402)
# https://three.ws/.well-known/openapi.yaml HTTP API spec
Sitemap: https://three.ws/sitemap.xml