/ developers
Build on Inferrex.
One API, one SDK, one MCP server — over the same self-healing integration layer the platform runs on. Point it at an API, it infers the schema; you read clean, governed data back.
/ three ways in
Pick how you talk to it.
Every path hits the same REST API at api.inferrex.com and is scoped to your workspace by your API key. Use whichever fits how you build.
REST API
Plain HTTPS + JSON at `https://api.inferrex.com`. Authenticate with an `X-API-Key` header. Everything the platform does is an endpoint.
TypeScript SDK
`@inferrex/sdk` wraps every endpoint in a typed client — `new InferrexClient({ apiKey })` — so you get autocomplete instead of curl.
MCP server
`@inferrex/mcp` exposes the same operations as Model Context Protocol tools, so Claude, Cursor and other agents can connect an API and read data for you.
What's available today.
@inferrex/sdk npm package and the hosted mcp.inferrex.com endpoint publish at launch — the examples throughout these docs show the exact shape you'll use.Authenticate with an API key.
One header, scoped to your workspace.
Every request carries an X-API-Key header:
curl https://api.inferrex.com/v1/connections \
-H "X-API-Key: $INFERREX_API_KEY"Create your first key in the dashboard — app.inferrex.com → Settings → API keys. The full key is shown once; store it somewhere safe (an environment variable, a secrets manager — never in source). After that you can mint further scoped keys programmatically with client.settings.createApiKey(...), and revoke any of them from the same place.
Keys are workspace-scoped: a key only ever sees its own workspace's connections, pipelines and data. There is no global or cross-tenant key.
Self-hosting & deployment.
The same API, wherever it runs.
The API surface is identical across every deployment topology — your code doesn't change between Cloud and air-gapped:
Cloud
Hosted multi-tenant at `api.inferrex.com`. Nothing to run; just an API key.
Enterprise
The same containers in your private VPC, your own base URL. Point the SDK at it with `new InferrexClient({ apiKey, baseUrl })`.
Sovereign
Self-hosted on your Kubernetes with bundled AI weights and the connector library. Offline licence, zero egress.
Air-gapped
No internet at all. Bundled models, pre-loaded library, signed offline licence.
The deployment modes are covered in depth on the technical page.