test_mcp_server
A simple MCP (Model Context Protocol) server project demonstrating both local and remote MCP server setups using FastMCP, LangChain, and uv.
Requirements
- Python 3.9+
pipuv
First-Time Setup
1. Install uv
pip install uv
2. Navigate to the project directory
cd test_mcp_server
3. Initialize the project with uv
uv init .
4. Add FastMCP
uv add fastmcp
Local MCP Server Setup
1. Create the local server file
Create a file named:
local_server.py
This file contains your MCP server implementation.
2. Add required dependencies
uv add langchain langchain-openai langchain_mcp_adapters
3. Create the client
Create a client file:
client.py
4. Run the local server using STDIO
The local MCP server communicates via STDIO.
Run the client:
uv run client.py
Remote MCP Server Setup
1. Create MCP tools
- Define your MCP tools for the remote server
- Ensure they are compatible with FastMCP Cloud
2. Deploy to FastMCP Cloud
- Deploy the server to FastMCP Cloud
- Obtain the remote server configuration
3. Update configuration
- Add the remote MCP server configuration to your config file
- Replace the local STDIO setup with the remote server endpoint
4. Run the client with the remote server
uv run client.py
Notes
- Local server uses STDIO for communication
- Remote server runs on FastMCP Cloud
uvhandles dependency management and execution- Same client can be used for both local and remote servers by changing configuration
Remote MCP Server Deployment (FastMCP Cloud)
GitHub Access
- GitHub repository access was granted to FastMCP Cloud
- FastMCP Cloud pulls the source code directly from the repository
Deployment Steps
- Created MCP tools for the remote server
- Connected the GitHub repository to FastMCP Cloud
- FastMCP Cloud executed the server using:
main.py
- The server was successfully deployed as a remote MCP server
Client Configuration
- Updated the MCP configuration file to point to the remote FastMCP Cloud endpoint
- Reused the same
client.pyfor both local and remote execution
Running the Client
uv run client.py
