Use OpenMind from your own code
OpenMind exposes a read-only public REST API at `/api/public/v1/...` so you can pull projects, conversations, messages, nodes, and edges into whatever you're building. Auth is per-user personal-access tokens; same RLS scope as your browser session.
Mint a token
Settings → API keys → Create. The token is shown once. Copy it the moment it appears; there's no way to reveal it again — only to revoke and mint a new one. Tokens are scoped to your user, not to a project, so they see exactly the rows your browser sees.
Make a request
`curl -H 'Authorization: Bearer <token>' https://api.openmind.app/api/public/v1/projects` returns the same JSON your projects-list page renders. Every endpoint is RLS-gated; a token belonging to alice will not see bob's rows even if a malicious caller guesses bob's project ID.
What's in scope
Today: projects, conversations, messages, nodes, edges, exports — everything readable via the dashboard. Out of scope: any write verb (PATCH / POST / DELETE) and admin endpoints. Write access is on the roadmap behind a separate scope; revisit when we ship it.
Rate limits + cost
The hosted demo applies a soft per-token throttle (60 req/min); self-host has no built-in rate limit. LLM-backed endpoints (`/ask`, extraction triggers) bill against whichever provider key signs the request — your own if you've added one in Settings → API keys, the operator's otherwise.
The full schema (every endpoint, every field, every error envelope) lives at https://api.openmind.app/docs — FastAPI's auto-generated OpenAPI page, which doubles as a Swagger UI you can poke from the browser.