Connect a source, let the Brain build, point your agent at it. The reference below covers the happy path; for anything deeper, we onboard every new team directly.
Create an org in the app and generate a key under API Keys. Keys are org-scoped and shown once at creation.
The CLI extracts schema metadata locally — only structure is sent to BrainBox, never your rows and never a connection string.
npm install -g @brainbox-labs/cli brainbox repo init # link this project to your brain
Start with one database or a folder of docs. Each additional source enriches the same brain.
brainbox meta ingest db # schema metadata from a database brainbox meta ingest doc # definitions & policy from documents
The Brain is served as a hosted MCP server. Add it to Claude, Cursor, or your own orchestrator and the grounded tools appear automatically.
# MCP endpoint (authenticate with your API key) https://ctx.brainbox-ai.app/api/mcp
Building something more involved — an embedded multi-tenant setup, a custom connector, the SDK? Start from the developer platform or talk to us — we onboard every new team directly.
Book a demo and we'll build a brain from a sample of your stack.