Tushare MCP Server

English | 中文

This project is a Model Context Protocol (MCP) server based on the Tushare data interface. It provides rich and comprehensive financial market data for mainland China, designed to be used with AI assistants.

🌟 Features

• Stock Data: Access to basic stock information, daily/weekly/monthly quotes, and historical data.
• Market Data: Includes Top List (Longhubang), Margin Trading, Block Trades, and Money Flow data.
• Financial Data: Retrieve financial statements and key indicators for listed companies.
• Index Data: Get quotes and historical data for major market indices.
• Fund & ETF Data: Access to ETF quotes, fund NAV, holdings, and ratings.
• Bond Data: Provides basic information and market data for bonds and convertible bonds.
• Macroeconomics: Offers data on GDP, CPI, interest rates, exchange rates, PMI, and money supply.
• Smart Search: Search for stocks by name or code.
• High Performance: Built with asynchronous processing to support efficient queries.

📦 Installation

1. Clone the repository:

   git clone https://github.com/your-username/tushare_mcp.git
   cd tushare_mcp
   

2. Install the required dependencies:

   pip install -r requirements.txt
   

⚙️ Configuration

To use this server, you need to provide a Tushare API Token. You can get a free token by registering on the Tushare Pro website.

There are two ways to configure your token:

Method 1: Environment Variable (Recommended)

This is the most secure and flexible method. The server is designed to automatically read the token from an environment variable named TUSHARE_TOKEN.

• On macOS/Linux:

  export TUSHARE_TOKEN="your_tushare_token_here"
  

  To make this setting permanent, add the line to your shell's profile file (e.g., ~/.bash_profile, ~/.zshrc).

• On Windows:

  set TUSHARE_TOKEN="your_tushare_token_here"
  

  To set it permanently, you can use the System Properties > Environment Variables panel.

Method 2: Configuration File (for MCP Clients)

If you are using an MCP client that launches the server for you (like Claude Desktop), you should use the mcp_config.json file.

1. Create the config file by copying the template:

   cp mcp_config.json.template mcp_config.json
   

2. Edit mcp_config.json: Open the file and replace the placeholder values.

   • cwd: Set this to the absolute path of the tushare_mcp project directory.
   • TUSHARE_TOKEN: Paste your Tushare token here.

   Example mcp_config.json:

   {
     "mcpServers": {
       "tushare": {
         "command": "python",
         "args": ["server.py"],
         "cwd": "/path/to/your/tushare_mcp",
         "env": {
           "TUSHARE_TOKEN": "YOUR_TUSHARE_TOKEN_HERE"
         }
       }
     }
   }
   

Note: If the TUSHARE_TOKEN is not set by any of these methods, the server will fall back to the free, non-Pro version of the Tushare API, which has significant limitations on data access.

▶️ Usage

There are two primary ways to run the server:

1. Standalone Mode

You can run the server directly from your terminal. This is useful for testing or for connecting with clients that attach to a running process.

1. Set your API token using the environment variable method described above.

2. Start the server:

   python server.py
   

3. Verify: Upon successful startup, you will see one of the following messages:

   • ✅ Tushare Token has been set. (If you configured a token)
   • ⚠️ Tushare Token is not set, using free interface (limited functionality). (If no token is found)

   The server is now running and waiting for a client to connect.

2. With an MCP Client (e.g., Claude Desktop)

Most MCP clients will manage the server process for you.

1. Configure mcp_config.json as described in the configuration section, making sure the cwd and TUSHARE_TOKEN are set correctly.
2. Point your client to this mcp_config.json file.
3. The client will automatically start the server.py process when you interact with one of its tools. You do not need to run python server.py manually.

🛠️ Available Tools

This server exposes a comprehensive set of tools to access financial data. Here are some of the categories:

• Basic Stock Data: get_stock_basic, get_daily_data, get_weekly_data, get_monthly_data, etc.
• Market Reference Data: get_top_list, get_money_flow, get_margin_detail, get_block_trade.
• Financial Data: get_financial_data, get_balance_sheet, get_cash_flow, get_fina_indicator.
• Index Data: get_index_data, get_index_basic, get_index_weight.
• ETF & Fund Data: get_etf_basic, get_etf_daily, get_fund_daily, get_fund_portfolio.
• Bond Data: get_bond_basic, get_bond_daily, get_cb_basic, get_cb_daily.
• Macroeconomic Data: get_gdp_data, get_cpi_data, get_interest_rate, get_exchange_rate.

For a complete list of tools and their parameters, please refer to the handle_list_tools function in server.py.

📄 License

This project is open-sourced under the GPL v3.0 License.

---

Tushare MCP 服务器

English | 中文

本项目是一个基于 Tushare 数据接口的模型上下文协议 (MCP) 服务器，为 AI 助手提供丰富、全面的中国内地金融市场数据。

🌟 功能特性

• 股票数据: 提供股票基本信息、日线/周线/月线行情及历史数据。
• 市场数据: 包含龙虎榜、融资融券、大宗交易、资金流向等数据。
• 财务数据: 获取上市公司的财务报表和关键财务指标。
• 指数数据: 获取主要市场指数的行情和历史数据。
• 基金与ETF: 提供 ETF 行情、基金净值、持仓、评级等信息。
• 债券数据: 提供债券和可转债的基本信息及行情数据。
• 宏观经济: 提供 GDP、CPI、利率、汇率、PMI、货币供应量等数据。
• 智能搜索: 支持按名称或代码搜索股票。
• 高性能: 采用异步处理，支持高效查询。

📦 安装指南

1. 克隆代码仓库：

   git clone https://github.com/your-username/tushare_mcp.git
   cd tushare_mcp
   

2. 安装所需依赖：

   pip install -r requirements.txt
   

⚙️ 配置说明

1. 您需要一个 Tushare API 令牌。请访问 Tushare Pro 官网 注册账户以获取您的令牌。

2. 将 Tushare 令牌设置为环境变量。这是推荐的最安全方式。

   • 在 macOS/Linux 系统中：

     export TUSHARE_TOKEN="your_tushare_token_here"
     

   • 在 Windows 系统中：

     set TUSHARE_TOKEN="your_tushare_token_here"
     

3. 或者，您也可以在 mcp_config.json 文件中配置（不推荐）。复制模板文件并填入您的信息：

   cp mcp_config.json.template mcp_config.json
   

   然后编辑 mcp_config.json，填入您的令牌和正确的项目路径。

▶️ 如何使用

通过以下命令运行服务器：

python server.py

服务器将启动并等待来自 MCP 客户端（例如 Claude Desktop）的连接。

如果未设置 TUSHARE_TOKEN，服务器将使用免费接口，其功能会受到限制。

🛠️ 可用工具

该服务器提供了一套全面的工具集，用于访问金融数据。以下是部分工具类别：

• 基础股票数据: get_stock_basic, get_daily_data, get_weekly_data, get_monthly_data 等。
• 市场参考数据: get_top_list, get_money_flow, get_margin_detail, get_block_trade。
• 财务数据: get_financial_data, get_balance_sheet, get_cash_flow, get_fina_indicator。
• 指数数据: get_index_data, get_index_basic, get_index_weight。
• ETF与基金数据: get_etf_basic, get_etf_daily, get_fund_daily, get_fund_portfolio。
• 债券数据: get_bond_basic, get_bond_daily, get_cb_basic, get_cb_daily。
• 宏观经济数据: get_gdp_data, get_cpi_data, get_interest_rate, get_exchange_rate。

关于完整的工具列表及其参数，请参考 server.py 文件中的 handle_list_tools 函数。

📄 开源协议

本项目采用 MIT 许可证 开源。