mcp-builder:deploy

Deploy MCP Server

You are helping the user deploy their MCP server to production.

Deployment Options

Platform	Best For	Complexity	Cost
Google Cloud Run	Auto-scaling, enterprise	Medium	Pay-per-use
Render	Simple deploys, small teams	Low	$7+/month
Fly.io	Edge deployment, low latency	Medium	Pay-per-use

Workflow

Phase 1: Choose Platform

Ask the user:

1. What's your expected traffic volume?
2. Do you need edge deployment (low latency globally)?
3. What's your budget?
4. Do you already use a cloud provider?

Phase 2: Prepare for Deployment

Verify server is production-ready:

• Environment variables externalized
• No hardcoded secrets
• Health check endpoint exists
• Logging configured
• Error handling complete

---

Google Cloud Run

Dockerfile

# Python FastMCP Server
FROM python:3.11-slim

WORKDIR /app

# Install dependencies
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

# Copy application
COPY . .

# Cloud Run sets PORT environment variable
ENV PORT=8080
EXPOSE 8080

# Start server
CMD ["python", "server.py"]

requirements.txt

fastmcp>=0.1.0
uvicorn>=0.30.0
python-dotenv>=1.0.0
httpx>=0.27.0
# Add your other dependencies

Deploy Script

#!/bin/bash
# deploy-cloud-run.sh

PROJECT_ID="your-project-id"
REGION="us-central1"
SERVICE_NAME="your-mcp-server"

# Build and push image
gcloud builds submit --tag gcr.io/$PROJECT_ID/$SERVICE_NAME

# Deploy to Cloud Run
gcloud run deploy $SERVICE_NAME \
  --image gcr.io/$PROJECT_ID/$SERVICE_NAME \
  --platform managed \
  --region $REGION \
  --allow-unauthenticated \
  --set-env-vars "NODE_ENV=production" \
  --memory 512Mi \
  --cpu 1 \
  --min-instances 0 \
  --max-instances 10

# Get the URL
gcloud run services describe $SERVICE_NAME \
  --region $REGION \
  --format 'value(status.url)'

Cloud Run Tips

• Set --min-instances 1 to avoid cold starts
• Use --cpu-boost for faster cold starts
• Set --concurrency based on your server's capacity
• Disable HTTP/2 if you have issues: --use-http2=false

---

Render

render.yaml

services:
  - type: web
    name: your-mcp-server
    runtime: python
    buildCommand: pip install -r requirements.txt
    startCommand: python server.py
    envVars:
      - key: PORT
        value: 10000
      - key: NODE_ENV
        value: production
      - key: PYTHON_VERSION
        value: 3.11.0
    healthCheckPath: /health
    autoDeploy: true

Deploy Steps

1. Push code to GitHub
2. Connect repo to Render
3. Render auto-detects render.yaml
4. Set environment variables in dashboard
5. Deploy

Render Tips

• Use "Web Service" type, not "Background Worker"
• Health check keeps service awake
• Auto-deploy on push to main branch

---

Fly.io

fly.toml

app = "your-mcp-server"
primary_region = "iad"

[build]
  dockerfile = "Dockerfile"

[http_service]
  internal_port = 8080
  force_https = true
  auto_stop_machines = true
  auto_start_machines = true
  min_machines_running = 0

[[vm]]
  cpu_kind = "shared"
  cpus = 1
  memory_mb = 512

Dockerfile (same as Cloud Run)

FROM python:3.11-slim
WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
ENV PORT=8080
EXPOSE 8080
CMD ["python", "server.py"]

Deploy Script

#!/bin/bash
# deploy-fly.sh

# Install Fly CLI if needed
# curl -L https://fly.io/install.sh | sh

# Login
fly auth login

# Create app (first time only)
fly apps create your-mcp-server

# Set secrets
fly secrets set AUTH0_DOMAIN=your-domain.auth0.com
fly secrets set AUTH0_AUDIENCE=https://your-api.com

# Deploy
fly deploy

# Check status
fly status

Fly.io Tips

• Use --ha for high availability (2+ machines)
• Deploy to multiple regions for global low latency
• fly scale count 2 for redundancy

---

Environment Variables

Set these in your deployment:

# Required
PORT=8080
NODE_ENV=production

# If using Auth
AUTH0_DOMAIN=your-tenant.auth0.com
AUTH0_AUDIENCE=https://your-api.com

# If using database
DATABASE_URL=postgresql://user:pass@host:5432/db

# Your API keys (keep secret!)
OPENAI_API_KEY=sk-...

---

Post-Deployment Checklist

• Health endpoint responds: curl https://your-server.com/health
• Tools list correctly: curl -X POST https://your-server.com/mcp ...
• Test one tool call end-to-end
• Configure custom domain (optional)
• Set up monitoring/alerts
• Test with real MCP client (Claude Desktop, Cursor)

Update State

After deployment, update .mcp-builder/state.json:

{
  "deployment": {
    "platform": "cloud-run",
    "url": "https://your-server-xyz.run.app",
    "region": "us-central1",
    "deployedAt": "2025-01-15T12:00:00Z"
  }
}