A Model Context Protocol (MCP) server that provides AI assistants with access to Shioaji trading API for the Taiwanese financial market.

This server implements the MCP protocol to expose Shioaji API functionality as tools that can be used by AI assistants. It allows AI models to:

• Retrieve current stock prices
• Fetch historical data
• List available stocks
• And more...

• Python 3.10 or higher
• uv (fast Python package manager)

uv sync

Before running the server, you need to configure your Shioaji API credentials. There are two ways to do this:

Set the following environment variables:

export SHIOAJI_API_KEY="your_api_key"
export SHIOAJI_SECRET_KEY="your_secret_key"

Create a .env file in the root directory with the following content:

SHIOAJI_API_KEY=your_api_key
SHIOAJI_SECRET_KEY=your_secret_key

Start the server with:

uv run mcp-server-shioaji

The server will start on http://0.0.0.0:8000 by default.

The server exposes the following tools via MCP:

Get the current price of a stock by its symbol.

{
  "tool": "get_stock_price",
  "params": {
    "symbols": "TW.2330,TW.2317"
  }
}

Response will include price information for the requested stocks, including open, high, low, close prices, volume, and other trading data.

Fetch K-Bar (candlestick) data for a stock within a date range.

{
  "tool": "get_kbars",
  "params": {
    "symbol": "TW.2330",
    "start_date": "2023-12-01",
    "end_date": "2023-12-15"
  }
}

If start_date is not provided, it defaults to today. If end_date is not provided, it defaults to the same as start_date.

Scan stocks based on various ranking criteria.

{
  "tool": "scan_stocks",
  "params": {
    "scanner_type": "VolumeRank",
    "ascending": false,
    "limit": 10
  }
}

Supported scanner types:

• VolumeRank - Ranking by trading volume
• AmountRank - Ranking by trading amount
• TickCountRank - Ranking by number of transactions
• ChangePercentRank - Ranking by percentage change
• ChangePriceRank - Ranking by price change
• DayRangeRank - Ranking by daily range

Default limit is 20, and results are sorted in descending order by default (set ascending to true for ascending order).

mcp-server-shioaji/
├── src/
│   └── mcp_server_shioaji/
│       ├── __init__.py      # Package entry point
│       └── server.py        # MCP server implementation
├── pyproject.toml           # Project metadata and dependencies
└── README.md                # This file

To add new Shioaji functionality, modify server.py and add new tool definitions using the @mcp.tool decorator.

MIT

• Shioaji - The Python wrapper for SinoPac's trading API
• MCP - Model Context Protocol