Llm MCP Server Template

🚀 llm-mcp-server-template

A template project for developing Large Language Model (LLM) MCP servers.

• The server is responsible for implementing the digital size comparison capability.
• The client supports multiple models and invocation methods.

🚀 Quick Start

MCP Introduction

• Official documentation: https://modelcontextprotocol.io/introduction
• A simple understanding:
  • The model (client) acts as the brain, responsible for planning and executing text-based tasks. For example, the first step is to write code, the second step is to execute it, and the third step is to analyze the results.
  • MCP (Server) serves as a tool, providing other capabilities required by the model, such as mathematical calculations, file reading and writing, and network requests.

✨ Features

MCP - Server

• Mainly implements MCP functions, such as mathematical calculations, file reading and writing, and network requests.
• Reference documentation: https://modelcontextprotocol.io/quickstart/server
• During development, you can use the mcp - server - inspector tool to assist with debugging.

Common Server Types:

1. Local Stdio, where the client directly calls the local server file (Python or Node.js).
   • Advantage: Push the Node project to the remote npm repository and run it via npx to achieve a semi - remote invocation effect.
2. Remote aggregation platforms using SSE, such as Glama.ai and MCP.so.

MCP - Client

• Responsible for coordinating the model and the MCP server, ultimately achieving the goal of "the model invokes MCP capabilities".

Common Client Types:

1. Use ready - made client software, such as Cursor and MCP Inspector.
2. Various intelligent agents on cloud platforms, such as Alibaba Cloud's Bailing application management.
3. Write scripts to support it yourself:
   • 3.1 If the model SDK already supports MCP, use scripts for convenient invocation:
     • Example: Claude (https://modelcontextprotocol.io/quickstart/client)
     • Example: OpenAI (https://openai.github.io/openai - agents - python/mcp/)
   • 3.2 If the model SDK does not support MCP, you need to implement it yourself:
     • 3.2.1 Use a framework, such as the LangChain - MCP adapter (https://github.com/langchain - ai/langchain - mcp - adapters).
     • 3.2.2 Implement it completely custom - made, specifically including multi - round requests:
       • a. Use a prompt to request the model, inform it of the task and the introduction of the MCP server, and the model will return the invocation interface and parameters.
       • b. Use the information returned in the previous round to request the tools and obtain the results.
       • c. Use the information returned in the previous round to request the model and obtain the results.
       • d. Repeat steps b and c until the model returns the final result.

📦 Installation

Environment Preparation

# Install the uv package management tool (skip if already installed)
curl -LsSf https://astral.sh/uv/install.sh | sh
source ~/.local/bin/env
uv --version

# Install Python (skip if already installed)
python --version

# Install pip (skip if already installed)
pip install --user --upgrade pip

# Install project dependencies
pip install -r requirements.txt

📚 Documentation

Project Results

The following is a brief description of the project results:

• MCP Server Debugging Interface
• MCP Client Invocation Example

TODO

The following features are not yet completed: 7. Package the MCP client and server with fixed functions into a desktop application using Electron.