legacy-client-mcp-server

内容来源：README.md（说明文档） · 原始地址 · 查看安装指南

原始内容

# GitHub Repository Assistant

A tool that uses the MCP (Model Control Protocol) to connect Claude with GitHub repositories, allowing users to analyze code and get assistance with tasks like creating deployment workflows.

## Features

- Clone and analyze any GitHub repository
- Interact with repositories through Claude 3.5 Sonnet
- Request deployment workflows and other code generation tasks
- Seamless integration via Streamlit frontend
- MCP server with specialized tools for repository interaction

## Requirements

- Python 3.13 or higher
- Git installed and available in PATH
- An Anthropic API key

## Installation

1. Clone this repository:
   ```
   git clone <your-repo-url>
   cd <repo-directory>
   ```

2. Install the required dependencies:
   ```
   pip install -r requirements.txt
   ```

3. Set your Anthropic API key in the `.env` file:
   ```
   ANTHROPIC_API_KEY=your_api_key_here
   ```

## Usage

1. Start the Streamlit application:
   ```
   streamlit run app.py
   ```

2. In the sidebar, connect to the MCP server by entering the path to the `server.py` file in the "Server Script Path" field and clicking "Connect to Server".

3. Enter a GitHub repository URL (e.g., `https://github.com/deploypilotorg/example-repo`) and click "Analyze Repository".

4. Once the repository is analyzed, you can interact with it by asking questions or requesting tasks like:
   - "What files are in this repository?"
   - "What does the main.py file do?"
   - "Create a GitHub Actions workflow for deploying this code"
   - "Generate a Dockerfile for this application"

## How It Works

The application uses the Model Control Protocol (MCP) to enable Claude to interact with GitHub repositories through specialized tools:

1. The MCP server (`server.py`) provides tools for cloning, analyzing, and reading from GitHub repositories
2. The Streamlit frontend (`app.py`) provides a user-friendly interface for interactions
3. Claude uses the provided tools to understand repository structure and generate relevant responses

## Limitations

- Currently only supports public GitHub repositories
- Large repositories may take longer to clone and analyze
- Generated deployment workflows may need manual adjustments based on specific requirements

GitHub Repository Assistant

A tool that uses the MCP (Model Control Protocol) to connect Claude with GitHub repositories, allowing users to analyze code and get assistance with tasks like creating deployment workflows.

Features

Clone and analyze any GitHub repository
Interact with repositories through Claude 3.5 Sonnet
Request deployment workflows and other code generation tasks
Seamless integration via Streamlit frontend
MCP server with specialized tools for repository interaction

Requirements

Python 3.13 or higher
Git installed and available in PATH
An Anthropic API key

Installation

Clone this repository:

git clone <your-repo-url>
cd <repo-directory>

Install the required dependencies:
```
pip install -r requirements.txt
```
Set your Anthropic API key in the .env file:
```
ANTHROPIC_API_KEY=your_api_key_here
```

Usage

Start the Streamlit application:
```
streamlit run app.py
```
In the sidebar, connect to the MCP server by entering the path to the server.py file in the "Server Script Path" field and clicking "Connect to Server".
Enter a GitHub repository URL (e.g., https://github.com/deploypilotorg/example-repo) and click "Analyze Repository".
Once the repository is analyzed, you can interact with it by asking questions or requesting tasks like:
- "What files are in this repository?"
- "What does the main.py file do?"
- "Create a GitHub Actions workflow for deploying this code"
- "Generate a Dockerfile for this application"

How It Works

The application uses the Model Control Protocol (MCP) to enable Claude to interact with GitHub repositories through specialized tools:

The MCP server (server.py) provides tools for cloning, analyzing, and reading from GitHub repositories
The Streamlit frontend (app.py) provides a user-friendly interface for interactions
Claude uses the provided tools to understand repository structure and generate relevant responses

Limitations

Currently only supports public GitHub repositories
Large repositories may take longer to clone and analyze
Generated deployment workflows may need manual adjustments based on specific requirements