This guide is for local development. To deploy on a server, see Docker Compose.
Prerequisites
- Node.js 20+ and pnpm —
npm install -g pnpm - Docker — used for Redis only
- A Supabase project — free tier works
- An Anthropic API key
Step 1: Clone and install
Step 2: Configure environment
.env and fill in:
Step 3: Run the database schema
Open your Supabase project → SQL Editor → paste and run the full contents ofsupabase/schema.sql.
Step 4: Start Redis
localhost:6379. The API and worker need it to run.
Step 5: Start the app
| Service | URL | What it does |
|---|---|---|
| API | http://localhost:3000 | REST API, identity resolution, signal ingestion |
| Frontend | http://localhost:5173 | Web UI |
| MCP server | http://localhost:3001 | Agent tool interface |
| Worker | background | AI synthesis, live sync, ICP scoring |
Step 6: Connect your first integration
Openhttp://localhost:5173 → Settings → Integrations and connect at least one source:
HubSpot
One-time import of existing contacts and deal history.
Gmail
Ingest email signals in real time via OAuth.
Pull connection requests and messages via Unipile.
Webhooks
Connect any custom signal source with HMAC auth.
Step 7: Connect your AI agent via MCP
Add Nous to yourmcp.json (Claude Desktop, Cursor, or any MCP host):
What’s next
How it works
Identity resolution, signal ingestion, and the unified record architecture.
MCP tools reference
Full reference for all MCP tools your agents can call.
Production deploy
Deploy on your own server with Docker Compose.
REST API
Integrate via REST API from any language or platform.