Kaggle MCP Server

kaggle mcp服务器

该项目提供了一个MCP（模型上下文协议）服务器，用于从Claude Desktop（或任何兼容MCP的客户端）与Kaggle进行交互

🚀 设置

1.克隆存储库并安装依赖项

 pip install -r requirements.txt

2.配置Kaggle API凭据

  Go to your Kaggle account -> Settings
  
  Scroll to API section → click Create New API Token.
  This will download a file called kaggle.json.

  Place kaggle.json in the following location on your system:
  
  Linux/Mac:
  
  ~/.kaggle/kaggle.json
  
  Windows:
  
  C:\Users\\.kaggle\kaggle.json

3.配置克劳德桌面

 To connect Claude with this MCP server, create or update the file:
 
 Claude/claude_desktop_config.json
 
 with the following contents:
 
 {
   "mcpServers": {
     "Kaggle": {
       "command": "
",
       "args": ["
"]
     }
   }
 }
 
 Replace 
 with the path to your Python interpreter (e.g. python or C:/Python/python.exe).
 
 Replace 
 with the path to this repo’s kaggle-server.py file.

📜 用法

 Once configured:
 
 Start Claude Desktop.
 
 Start the Kaggle MCP server (Run kaggle-server.py as a Python File).
 
 You can now use MCP tools defined in kaggle-server.py, for example:
 
 get_competitions_list: Fetches the list of available Kaggle competitions and returns them in a JSON-friendly format.

🛠️ 关于代码

 kaggle-server.py registers MCP tools using the @mcp.tool() decorator.
 
 The server communicates with Claude via MCP and exposes Kaggle API methods.
 
 Example tool:
 
 @mcp.tool()
 def get_competitions_list() -> list[dict]:
     """
     Fetch the list of available Kaggle competitions.
     """

✅ 需求

  Python 3.10+
      
  Kaggle API credentials (kaggle.json)
      
  Dependencies listed in requirements.txt

🔒 安全说明

 Never commit your kaggle.json file.
 
 Ensure the .kaggle folder has restricted permissions.