Lead Stampede MCP Server

The live backend that powers Lead Stampede Agent Cards. Exposes SMB business data as tools that AI agents can call over HTTP.

What this is

An Express server that implements four MCP tools:

Each request is authenticated via an X-Agency-API-Key header that maps to a row in the agencies table in Supabase. Clients are scoped to their agency, so one agency can never read another's data.

Setup

1. Install

npm install

2. Configure environment

cp .env.example .env

Then edit .env and fill in:

• SUPABASE_URL — your project URL (e.g. https://xlhxnlhxeprtzwbyepza.supabase.co)

• SUPABASE_SERVICE_ROLE_KEY — from Supabase → Project Settings → API → service_role secret

• PORT — defaults to 3000

Never commit .env to git. It's already in .gitignore.

3. Run locally

npm run dev

You should see:

Lead Stampede MCP server listening on port 3000

Testing it works

Health check (no auth)

curl http://localhost:3000/health
# → {"status":"ok","timestamp":"..."}

List available tools (auth required)

curl -H "X-Agency-API-Key: YOUR_AGENCY_API_KEY" \
  http://localhost:3000/mcp/tools

Call a tool

curl -X POST \
  -H "X-Agency-API-Key: YOUR_AGENCY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"arguments":{"client_slug":"grandinetti-molinar-law"}}' \
  http://localhost:3000/mcp/tools/get_business_profile

Expected response:

{
  "tool": "get_business_profile",
  "result": {
    "business_name": "Grandinetti & Molinar Law",
    "description": "Austin-based law firm founded in 2002...",
    "industry": "legal",
    ...
  },
  "response_ms": 42
}

Deploying to Railway

1. Push this folder to a new GitHub repo.

2. In Railway: New Project → Deploy from GitHub repo.

3. Select this repo.

4. Under Variables, add:

   • SUPABASE_URL

   • SUPABASE_SERVICE_ROLE_KEY

5. Railway auto-detects Node.js, installs deps, and runs npm start.

6. Once deployed, Railway gives you a public URL like mcp-production.up.railway.app.

7. Point a custom subdomain (e.g. mcp.leadstampede.io) at it via Railway → Settings → Domains.

Architecture

AI agent  →  POST /mcp/tools/:name  →  auth.js → tools.js → supabase.js
                                         ↓
                                   analytics.js (fire-and-forget)
                                         ↓
                                   mcp_tool_calls table

Next steps

• Build the Cloudflare Worker that serves Agent Cards at agentcards.leadstampede.io/{slug}.

• Build the onboarding form that populates clients rows in Supabase.

• Build the agency portal dashboard.