Xinkuang China Stock MCP

What xinkuang-china-stock-mcp MCP server does

The xinkuang-china-stock-mcp MCP server gives MCP-compatible assistants access to a broad set of Chinese equity and market-data operations. It covers A-shares, B-shares, and Hong Kong stocks where the relevant tool supports them, plus index and industry-board data.

Available operations include historical and real-time quotations, stock news, balance sheets, income statements, cash-flow statements, financial metrics, fund-flow history, insider transactions, shareholder details, product or business composition, profit forecasts, dividends, research reports, management changes, restricted-share release schedules, valuations, and volatility calculations. The server also exposes stock-code lists, macroeconomic indicators, investor sentiment, index constituents, index history, technical screening results, and industry-board summaries.

Historical quotes can include more than 30 technical indicators, such as SMA, EMA, RSI, MACD, Bollinger Bands, ATR, OBV, and MFI. Results can be requested as JSON, CSV, XML, Excel, Markdown, or HTML, depending on the tool.

How it works

The xinkuang-china-stock-mcp MCP server is implemented around FastMCP. Its core server definitions are in server.py, while __main__.py provides the command-line entry point and runtime modes. It supports local stdio communication for desktop clients and an HTTP network mode for deployments that need a network service.

Underlying market retrieval uses the akshare-one library and connects to Eastmoney, direct Eastmoney endpoints, Sina Finance, and Xueqiu. A fallback mechanism tries alternate sources when a preferred source fails or returns no data. In-memory and disk caching are included to reduce repeated retrievals and support cached index information.

Most stock tools accept a security symbol and an output format. Historical-data requests additionally accept date ranges, interval settings, adjustment mode, and optional indicator lists. The volatility tool uses minute-level history and accepts a minute period. Macro-data requests use a named indicator such as GDP, CPI, PMI, money supply, or stock summary.

Setup and configuration

The repository documents three deployment paths: Smithery installation for Claude Desktop, Docker, and local source execution. A Docker deployment can use the published image and expose port 8081:

docker pull ghcr.io/xinkuang/china-stock-mcp:latest
docker run -p 8081:8081 ghcr.io/xinkuang/china-stock-mcp:latest

For local execution, the documented prerequisites are Python 3.12 or newer, Git, and uv as the recommended package manager. Clone the repository, run uv sync or install requirements.txt, then start the default stdio mode with uv run -m china_stock_mcp. The README also identifies HTTP mode as supported, although the excerpt does not provide its exact launch command or endpoint.

No API keys, tokens, or other credentials are specified in the provided setup instructions. Data-source availability may affect individual requests because the server depends on external market-data endpoints.

Limitations and notes

The material describes retrieval and analysis tools, not order execution or brokerage connectivity. It does not state service-level guarantees, historical-data completeness, exchange licensing terms, or the update frequency of each source. Symbols and date formats vary by tool: historical stock data uses YYYY-MM-DD, while some index operations use YYYYMMDD or YYYYMM.

The xinkuang-china-stock-mcp MCP server can return several output formats, but the default is generally Markdown. Technical-indicator requests recommend keeping the requested indicator count to 10 or fewer. The documented client examples mention Claude and Cursor; compatibility with other MCP clients is not established by the provided material.