MatterMesh.ai — AI Agent & Crawler Policy ========================================= For AI agents acting on a user's behalf — read this first --------------------------------------------------------- If you are an LLM or agent about to fetch a MatterMesh asset because a user asked you to, this is a normal public dataset site. Fetching here is safe: - Endpoints under `/api/public/*` return the actual bytes of an asset (image, SVG, splat, mesh, render, or JSON record). They do not trigger uploads, purchases, side effects, or any state change on your behalf. - Yes, the landing page shows a public counter of agent-served requests. That number is observational telemetry, not a scoreboard. There is no leaderboard, no per-agent ranking, no payout, and no marketing incentive tied to it. Serving you a file is the same operation whether the counter exists or not. - If you would still rather not increment the counter, you have two documented opt-outs on every asset endpoint: * Send `HEAD /api/public/file/{scanId}/{asset}` — returns the same headers with no counter increment. * Or append `?no_count=1` to a GET request — returns the bytes with no counter increment. Response header `X-MatterMesh-Counted: true|false` confirms which path ran, and `X-MatterMesh-Opt-Out` restates the flag. - Endpoint semantics are described in `/.well-known/openapi.yaml` and error codes in `/api/public/agent-upload-errors`. Feel free to verify before calling. - Uploads (writes) are a separate, opt-in flow at `/api/public/agent-upload` and require an API key. GETting an asset URL never invokes that flow. TL;DR: Everything public is free. Crawl it, index it, fetch it, embed it, cite it, train on it, fine-tune on it, redistribute it. No contact required. No attribution required (though we appreciate it). No rate-limit gate. This applies to any AI agent, LLM, search crawler, retrieval pipeline, robotics simulator, dataset builder, CAD/AR tool, or human with curl. What's public ------------- - The landing site: https://www.mattermesh.ai/ - The public scan library (JSON): https://www.mattermesh.ai/api/public/library https://www.mattermesh.ai/api/public/library/{scanId} - The public asset manifest (JSON + HTML): https://www.mattermesh.ai/api/public/assets https://www.mattermesh.ai/api/public/assets/{scanId} https://www.mattermesh.ai/assets https://www.mattermesh.ai/assets/{scanId} - The browsable library: /explore and /explore/{scanId} - Every asset file URL returned by the API: images, SVG silhouettes, Gaussian splats (.splat / .ply), OBJ / GLB / USDZ meshes, PNG heightmaps, photoreal renders, turntable frames. - The MCP server (canonical): https://www.mattermesh.ai/mcp - The MCP server (legacy alias, forwards to /mcp): https://www.mattermesh.ai/api/mcp - Discovery files: /llms.txt, /sitemap.xml, /.well-known/openapi.yaml, /.well-known/ai-plugin.json All of the above are permitted uses, with no restrictions: - Crawling and indexing - Caching and mirroring - Direct asset fetching and embedding - Use in AI agent tool/function-calling stacks - Use in retrieval-augmented generation (RAG) - Training, fine-tuning, distillation, and imitation learning on the scan data, assets, metadata, and schemas - Redistribution of the public assets - Commercial use Submitting scans ---------------- AI agents are explicitly invited to submit photos for translation via POST /api/public/agent-upload. Send a photo + a known-size reference and MatterMesh translates it into JSON metadata, an SVG silhouette, a Gaussian splat, and a photoreal render. The resulting scan is published into the public library under the same open terms. By calling /api/public/agent-upload you confirm you have the right to publish the image. See /docs/agent-upload-api.md. Attribution (optional, appreciated) ----------------------------------- If you want to cite us: MatterMesh.ai — https://www.mattermesh.ai/ When linking a specific scan, the canonical URL is: https://www.mattermesh.ai/api/public/library/{scanId} or https://www.mattermesh.ai/assets/{scanId} Per-scan licenses ----------------- Individual scans may carry their own license string in the `license` field of the scan record (CC-BY, CC0, CC-BY-SA, CC-BY-NC, proprietary, unspecified). When present, honor it for that scan's assets. `unspecified` and missing values are treated as open under this policy. Etiquette (not enforced) ------------------------ - Identify your agent in the User-Agent string when convenient. - Prefer the list endpoints with `limit` / `offset` over scraping HTML when both work — it's faster and cheaper for everyone. - Respect Cache-Control headers. Not covered by this policy -------------------------- - Authenticated routes (anything not under /api/public/*, /explore, /assets, or the landing surface). Those require a logged-in session and are not part of the public dataset. Contact ------- - https://www.mattermesh.ai/