FastMCP Patient Information Server

A FastMCP server that provides patient information retrieval functionality through the get_patient tool.

Features

• get_patient tool: Retrieves comprehensive patient information including:

  • Personal details (name, DOB, contact info)

  • Insurance plans (Medical, Dental, Vision)

  • Employee information

Prerequisites

• Docker and Docker Compose installed on your Ubuntu server

• Network access to the server from Watson Orchestrate

Installation & Deployment

1. Transfer Files to Ubuntu Server

Copy the entire fastmcp-server directory to your Ubuntu server:

scp -r fastmcp-server/ user@your-ubuntu-server:/path/to/deployment/

2. Build and Run with Docker Compose

SSH into your Ubuntu server and navigate to the directory:

ssh user@your-ubuntu-server
cd /path/to/deployment/fastmcp-server

Build and start the container:

docker-compose up -d --build

3. Verify the Server is Running

Check container status:

docker-compose ps

View logs:

docker-compose logs -f

4. Test the Server Locally

curl http://localhost:8002/health

Connecting to Watson Orchestrate

Step 1: Get Your Server URL

Your MCP server will be accessible at:

http://your-ubuntu-server-ip:8002

Or if you have a domain:

https://your-domain.com:8002

Important: For production, use HTTPS with proper SSL certificates.

Step 2: Access Watson Orchestrate Tools Page

1. Log in to your Watson Orchestrate instance

2. Navigate to Skills or Tools section

3. Click on Add Tool or Connect Tool

Step 3: Configure MCP Connection

1. Connection Type: Select "MCP Server" or "Custom API"

2. Server URL: Enter your server endpoint

   http://your-ubuntu-server-ip:8002

3. Authentication (if required):

   • Type: None (for basic setup)

   • For production, implement OAuth2 or API key authentication

4. Connection Name: Patient Information MCP Server

5. Description: Provides patient information retrieval capabilities

Step 4: Discover Tools

1. Click Discover Tools or Test Connection

2. Watson Orchestrate will query the MCP server for available tools

3. You should see the get_patient tool appear

Step 5: Configure the get_patient Tool

The tool will be automatically configured with:

• Tool Name: get_patient

• Description: Retrieve patient information by contact ID

• Parameters:

  • contact_id (optional string): The unique contact identifier

Step 6: Test the Tool

1. In Watson Orchestrate, select the get_patient tool

2. Test with the sample contact ID: 7c50f84d-62af-f011-bbd3-000d3a9b6dcb

3. Or test without parameters to get default patient data

4. Verify the response matches the expected format

Step 7: Create Skills Using the Tool

1. Navigate to Skills Builder in Watson Orchestrate

2. Create a new skill that uses the get_patient tool

3. Configure the skill flow and parameters

4. Test and publish the skill

Watson Orchestrate Connection YAML Example

If Watson Orchestrate supports YAML configuration, create a connection file:

apiVersion: v1
kind: Connection
metadata:
  name: patient-mcp-server
spec:
  type: mcp
  endpoint: http://your-ubuntu-server-ip:8002
  authentication:
    type: none
  tools:
    - name: get_patient
      enabled: true

Security Considerations for Production

1. Enable HTTPS

Use a reverse proxy (nginx) with SSL:

# Install nginx
sudo apt-get update
sudo apt-get install nginx certbot python3-certbot-nginx

# Configure SSL
sudo certbot --nginx -d your-domain.com

2. Add Authentication

Update server.py to include API key authentication:

from fastapi import Header, HTTPException

@mcp.tool()
def get_patient(contact_id: str = None, api_key: str = Header(None)) -> Dict[str, Any]:
    if api_key != "your-secret-api-key":
        raise HTTPException(status_code=401, detail="Invalid API key")
    # ... rest of the function

3. Network Security

• Use firewall rules to restrict access

• Consider VPN or private network connection

• Implement rate limiting

4. Update docker-compose.yml for Production

version: '3.8'

services:
  fastmcp-server:
    build: .
    container_name: patient-mcp-server
    ports:
      - "127.0.0.1:8002:8002"  # Only bind to localhost
    environment:
      - PYTHONUNBUFFERED=1
      - API_KEY=${API_KEY}  # Use environment variable
    restart: unless-stopped
    networks:
      - mcp-network

networks:
  mcp-network:
    driver: bridge

Troubleshooting

Server Not Starting

# Check logs
docker-compose logs

# Rebuild container
docker-compose down
docker-compose up -d --build

Connection Refused from Watson Orchestrate

1. Verify firewall allows port 8002

2. Check server is listening on 0.0.0.0, not 127.0.0.1

3. Test with curl from another machine

Tool Not Appearing in Watson Orchestrate

1. Verify MCP server is responding to tool discovery requests

2. Check Watson Orchestrate logs for connection errors

3. Ensure the server URL is correct and accessible

Management Commands

# Start the server
docker-compose up -d

# Stop the server
docker-compose down

# View logs
docker-compose logs -f

# Restart the server
docker-compose restart

# Update the server
git pull  # or copy new files
docker-compose up -d --build

Extending the Server

To add more tools, edit server.py:

@mcp.tool()
def get_patient_appointments(patient_id: str) -> Dict[str, Any]:
    """Get patient appointments"""
    # Implementation here
    pass

@mcp.tool()
def get_patient_claims(patient_id: str) -> Dict[str, Any]:
    """Get patient insurance claims"""
    # Implementation here
    pass

Then rebuild and restart:

docker-compose up -d --build

Support

For issues or questions:

• Check the logs: docker-compose logs

• Review FastMCP documentation: https://github.com/jlowin/fastmcp

• Review Watson Orchestrate MCP integration docs

License

[Your License Here]