This is the MCP server you deploy separately to Render.
It exposes one endpoint:
POST /mcp(JSON-RPC 2.0 over SSE)
And one health endpoint:
GET /health
scan_productanalyze_claimsgenerate_json
These names match the frontend ArmorIQ integration.
- Copy
.env.exampleto.env - Set
GEMINI_API_KEY - Install and start:
npm install
npm run startHealth check:
curl http://localhost:10000/health- Put this
mcp-serverfolder in a separate GitHub repo (recommended). - In Render:
- New -> Web Service
- Runtime:
Node - Build command:
npm ci - Start command:
npm run start - Root Directory: repo root (or set to folder containing this package.json)
- Add env vars in Render service:
NODE_ENV=productionPORT=10000MCP_SERVER_NAME=bitespy-main-mcpMCP_SERVER_VERSION=1.0.0GEMINI_API_KEY=<your gemini key>MCP_ALLOWED_ORIGIN=*MCP_MODEL=gemini-2.5-flash-lite
- Deploy and test:
GET https://<render-service>.onrender.com/healthPOST https://<render-service>.onrender.com/mcpwithinitializePOST https://<render-service>.onrender.com/mcpwithtools/list
- Register in ArmorIQ MCP registry:
- Name: same as
MCP_SERVER_NAME - URL:
https://<render-service>.onrender.com/mcp
{
"jsonrpc": "2.0",
"id": 1,
"method": "initialize",
"params": {
"protocolVersion": "2024-11-05",
"capabilities": {},
"clientInfo": { "name": "armoriq-agent", "version": "1.0.0" }
}
}{
"jsonrpc": "2.0",
"id": 2,
"method": "tools/list"
}{
"jsonrpc": "2.0",
"id": 3,
"method": "tools/call",
"params": {
"name": "scan_product",
"arguments": {
"prompt": "Analyze this product and return strict JSON",
"model": "gemini-2.5-flash-lite"
}
}
}