Quickstart
Go from zero to governed MLS queries in 5 minutes. This guide walks through the vendor integration path: register, get trust-scored, request MLS access, authenticate, and make your first governed query.
[i]Prerequisites
Deploy your MCP server
Register your server with Conduit via the CLI or API. This triggers a trust scan and assigns a gateway URL.
npm install -g conduit-cli
conduit login
conduit deploy https://your-mcp-server.com/mcpOn success, you receive a gateway URL and API key:
Server: your-mcp-server
Gateway: https://gateway.conduitapi.dev/s/your-org/your-mcp-server
API Key: cnd_live_xxxxxxxxxxxxxxxxxxxx
Trust: A (92/100)Get trust-scored
Conduit automatically scans your server on deploy. The scanner evaluates 5 dimensions: Security (40%), Compliance (20%), Auth (20%), Performance (10%), and Runtime (10%). You need a qualifying grade to request MLS access.
Trigger a rescan any time:
conduit scan your-mcp-serverSee Trust Scoring for the full scoring methodology.
Request MLS access
Browse the MLS directory and request access. The MLS administrator reviews your trust score and decides whether to approve your request.
POST /api/vendor/access-requests
{
"mls_id": "austin-mls",
"server_id": "your-server-id",
"use_case": "Property search for consumer app"
}Once approved, you can route queries to that MLS through the gateway.
Authenticate and query
Use your API key to make governed MCP requests through the gateway. The gateway enforces the MLS governance policy on every request.
curl -X POST https://gateway.conduitapi.dev/s/austin-mls/reso-feed \
-H "Authorization: Bearer cnd_live_xxxxxxxxxxxxxxxxxxxx" \
-H "Content-Type: application/json" \
-d '{
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"name": "search_properties",
"arguments": {
"city": "Austin",
"propertyType": "Residential",
"minPrice": 300000,
"maxPrice": 500000
}
},
"id": 1
}'Check governance metadata
Every response includes governance headers that your application should respect:
X-Conduit-Governed: true
X-Conduit-Anti-Training: true
X-Conduit-Stateless: true
X-RateLimit-Remaining: 58
X-RateLimit-Reset: 1707400000If X-Conduit-Anti-Training is true, the response data must not be used for model training, fine-tuning, or embedding generation.