Deployment

Production configuration and deployment patterns for GraphQL MCP servers.

Production Settings

import uvicorn
from graphql_mcp import GraphQLMCP

server = GraphQLMCP.from_api(
    api,
    name="Production API",
    graphql_http=False,  # Disable GraphiQL in production
    allow_mutations=True,
    auth=jwt_verifier    # Always use authentication
)

app = server.http_app(
    transport="streamable-http",
    stateless_http=True  # Required for load balancing
)

if __name__ == "__main__":
    uvicorn.run(
        app,
        host="0.0.0.0",
        port=8000,
        workers=4,
        log_level="info"
    )

Key settings:

• graphql_http=False — Disable GraphiQL and MCP Inspector in production
• stateless_http=True — No session state, safe for multiple workers and load balancers
• transport="streamable-http" — HTTP with streaming support

Docker

FROM python:3.11-slim

WORKDIR /app

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

COPY . .

EXPOSE 8000

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

requirements.txt:

graphql-mcp
graphql-api
uvicorn

Build and run:

docker build -t graphql-mcp-server .
docker run -p 8000:8000 \
  -e GRAPHQL_URL=https://api.example.com/graphql \
  -e GRAPHQL_TOKEN=your_token \
  graphql-mcp-server

Kubernetes

apiVersion: apps/v1
kind: Deployment
metadata:
  name: graphql-mcp
spec:
  replicas: 3
  selector:
    matchLabels:
      app: graphql-mcp
  template:
    metadata:
      labels:
        app: graphql-mcp
    spec:
      containers:
      - name: graphql-mcp
        image: your-registry/graphql-mcp:latest
        ports:
        - containerPort: 8000
        env:
        - name: GRAPHQL_URL
          valueFrom:
            secretKeyRef:
              name: graphql-secrets
              key: url
        - name: GRAPHQL_TOKEN
          valueFrom:
            secretKeyRef:
              name: graphql-secrets
              key: token
---
apiVersion: v1
kind: Service
metadata:
  name: graphql-mcp
spec:
  selector:
    app: graphql-mcp
  ports:
  - port: 80
    targetPort: 8000
  type: LoadBalancer

Use stateless_http=True in your server code when running with multiple replicas.

Serverless (AWS Lambda)

Deploy to AWS Lambda using Mangum:

from mangum import Mangum
from graphql_mcp import GraphQLMCP

server = GraphQLMCP.from_api(api, name="Lambda API")
app = server.http_app(stateless_http=True)

handler = Mangum(app)

Next Steps

• Customization — Auth, middleware, lifespan management
• API Reference — Full http_app parameter reference