fredycampino/mcp-server

The Fast and Secure Way to Build MCP Servers

This project provide a minimal mcp-server based on FastMCP framework that you can run in local. It's designed to be a robust starting point for any developer looking to expose custom tools to AI agents and LLMs.

Getting Started

1. Clone the repository

git clone git@github.com:fredycampino/mcp-server.git
cd mcp-server

If you're using HTTPS instead of SSH:

git clone https://github.com/fredycampino/mcp-server.git

2. Start the virtual environment

python3 -m venv .venv
source .venv/bin/activate   # On Linux/macOS
# Or, on Windows:
.venv\Scripts\activate

pip install -r requirements.txt

3. Run the server

python local_mcp.py

This will start the FastMCP server at http://127.0.0.1:7800/mcp using SSE transport.

Connecting with CLINE

To use this MCP server with CLINE, follow these steps:

1. Open CLINE.
2. Click the Manage MCP Servers icon located in the bottom-left corner of the chat interface.
3. Add or modify the configuration in the settings file (named cline_mcp_settings) with the following content:

{
  "mcpServers": {
    "local_mcp_server": {
      "disabled": false,
      "timeout": 60,
      "type": "sse",
      "url": "http://127.0.0.1:7800/mcp"
    }
  }
}

Connecting with LM Studio

To connect this MCP server to LM Studio, follow these steps:

1. Open LM Studio.
2. Open PROGRAM tab, in top rigth side.
3. Click Install and Open under the MCP section.
4. Opens and modify the mcp.json configuration file with the following content:

{
  "mcpServers": {
    "local_mcp_server": {
      "url": "http://127.0.0.1:7800/mcp"
    }
  }
}

Connecting with Cursor

To connect this MCP server to Cursor, follow these steps:

1. Open Cursor.
2. Go to Settings.
3. Click tools & Integrations.
4. Click in new MCP server and edit the mcp.json file:

{
  "mcpServers": {
    "local_mcp_server": {
      "url": "http://127.0.0.1:7800/mcp"
    }
  }
}

Add Your Own Tool Module

Extending the server with your own tools is straightforward. Here’s how to add a new module called x_tools.py:

Step 1: Create the tool file

Create a new file tools/x_tools.py. Define your functions inside it. Remember that they can be async or regular functions.

# in tools/x_tools.py

import asyncio

async def get_server_info():
    """A new tool to get server information."""
    # Simulate an async operation
    await asyncio.sleep(0.1) 
    return {"server_name": "MyMCP", "version": "1.0.0"}

def get_user_id():
    """A simple sync tool."""
    return {"user_id": "user-12345"}

Step 2: Register the new tools in local_mcp.py

Open local_mcp.py and import your new functions. Then, add them to the TOOLS list to be automatically registered when the server starts.

# in local_mcp.py

from fastmcp import FastMCP

# ... other imports
from tools.x_tools import get_server_info, get_user_id # 👈 Import your new functions

# ...

# List of imported tool callables to expose
TOOLS = [
    internet_research,
    get_url,
    # ... other tools
    get_server_info, # 👈 Add your new function
    get_user_id      # 👈 Add your other new function
]