ArturoHurtado7/MCP-Server

MCP-Server

Description

This project implements an MCP (Model Context Protocol) server in Python. The server exposes tools to obtain weather information for cities using the Open-Meteo API, and can be consumed by MCP clients or inspected with the MCP Inspector for VS Code.

Python dependencies installation

1. Create and activate a virtual environment (optional but recommended):

   python3 -m venv venv
   source venv/bin/activate
   

2. Install the project dependencies:

   pip install -r requirements.txt
   

Local development run

You can run the MCP server locally without Nginx or systemd, ideal for development and quick testing.

From the src folder, run:

../venv/bin/uvicorn main:app --host 0.0.0.0 --port 8000 --reload

This starts the server in autoreload mode (useful for development). Access it from your browser at:

http://localhost:8000/mcp/

Or from another machine on the network using your Raspberry Pi's IP:

http://192.168.10.101:8000/mcp/

Usage with MCP Inspector

You can use the MCP Inspector to debug and explore your MCP server from VS Code.

Recommended run with Inspector

From the project root, run:

npx -y @modelcontextprotocol/inspector /home/pi/Documents/Projects/MCP-Server/venv/bin/python /home/pi/Documents/Projects/MCP-Server/src/mcp_server/fast_mcp.py

Or if you use the main.py entrypoint:

npx -y @modelcontextprotocol/inspector /home/pi/Documents/Projects/MCP-Server/venv/bin/python /home/pi/Documents/Projects/MCP-Server/src/main.py

This will open the Inspector in your browser and connect to your MCP server using stdio.

Deployment as a service (daemon) with systemd and Nginx

1. Launch the MCP server as a systemd service

Create the file /etc/systemd/system/mcp-server.service with:

[Unit]
Description=MCP Server (uvicorn)
After=network.target

[Service]
User=pi
WorkingDirectory=/home/pi/Documents/Projects/MCP-Server/src
ExecStart=/home/pi/Documents/Projects/MCP-Server/venv/bin/uvicorn main:app --host 127.0.0.1 --port 8000
Restart=always

[Install]
WantedBy=multi-user.target

Then run:

sudo systemctl daemon-reload
sudo systemctl enable mcp-server
sudo systemctl start mcp-server

Check the status:

sudo systemctl status mcp-server

2. Nginx configuration as a reverse proxy

Install Nginx:

sudo apt update
sudo apt install nginx

Create the file /etc/nginx/sites-available/mcp with:

server {
    listen 80;
    server_name 192.168.10.101;

    location / {
        proxy_pass http://127.0.0.1:8000;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

Enable the configuration and reload Nginx:

sudo ln -s /etc/nginx/sites-available/mcp /etc/nginx/sites-enabled/
sudo nginx -t
sudo systemctl reload nginx

3. Access from the local network

With the server and Nginx running, access from any device on the network:

http://192.168.10.101/

4. Restart the service after code changes

sudo systemctl restart mcp-server

It is not necessary to restart Nginx after changes in the Python code.

Additional notes

• For production, use the deployment with systemd and Nginx.
• For development, use uvicorn with --reload.
• MCP Inspector is useful for debugging and interactive testing.