Trade-MCP

A modular trading automation project using Zerodha Kite Connect and MCP (Modular Command Platform) for tool-based and resource-based automation. This project exposes trading actions (like placing orders and fetching holdings) as MCP tools, making them accessible to Claude Desktop and other MCP-compatible clients.

Features

• Zerodha Integration: Place buy/sell orders and fetch holdings using the Kite Connect API.

• MCP Tools: Expose trading actions as callable tools for automation and AI agents.

• Environment-based Secrets: API keys and secrets are loaded from a .env file for security.

• Extensible: Easily add new tools or resources for more trading actions.

Related MCP server: Kite MCP Server

Setup

1. Clone the Repository

git clone <your-repo-url>
cd Trade-MCP

2. Create and Activate a Virtual Environment

uv venv
.venv\Scripts\activate  # On Windows
# or
source .venv/bin/activate  # On Linux/Mac

3. Install Dependencies

uv pip install -r requirements.txt
# or, if using pyproject.toml:
uv pip install .

4. Configure Environment Variables

Create a .env file in the project root:

API_KEY=your_zerodha_api_key
API_SECRET=your_zerodha_api_secret

5. Get Your Zerodha Access Token

To use the Zerodha API, you need an access token. Follow these steps:

1. Get Your API Key and Secret

   • Obtain these from the Zerodha developer console.

2. Generate the Login URL
   In a Python shell or script, run:

   from kiteconnect import KiteConnect
   kite = KiteConnect(api_key="your_api_key")
   print(kite.login_url())

   • Open the printed URL in your browser and log in with your Zerodha credentials.

3. Extract the request_token

   • After login, you will be redirected to your redirect URL (set in the Zerodha app settings).

   • The URL will look like:
     http://localhost:8000/?request_token=REQUEST_TOKEN_HERE&action=login&status=success

   • Copy the request_token value from the URL.

4. Exchange the request_token for an access_token
   In a Python shell or script, run:

   data = kite.generate_session("REQUEST_TOKEN_HERE", api_secret="your_api_secret")
   print(data["access_token"])

   • Replace "REQUEST_TOKEN_HERE" and "your_api_secret" with your actual values.

   • The printed value is your access_token.

5. Set the Access Token

   • You can set this as an environment variable or use it directly in your code as needed.

Usage

Run the MCP Server

uv run --with mcp[cli] mcp run server.py

Or, if MCP CLI is installed in your environment:

mcp run server.py

Claude Desktop Integration

1. Add a server entry in your claude_desktop_config.json:

   {
     "mcpServers": {
       "Trade-MCP": {
         "command": "uv.EXE", # Exact Path to uv executable
         "args": [
           "run",
           "--with",
           "mcp[cli]",
           "mcp",
           "run",
           "server.py" # Exact Path to your server script
         ]
       }
     }
   }

2. Restart Claude Desktop.

3. The tools will appear in Claude's tool list.

Exposed Tools

• add(a: int, b: int): Add two numbers.

• place_zerodha_buy_order(symbol: str, quantity: int): Place a buy order.

• place_zerodha_sell_order(symbol: str, quantity: int): Place a sell order.

• get_zerodha_holdings(): Get all holdings from Zerodha.

Extending

To add a new tool, define a function in server.py and decorate it with @mcp.tool():

@mcp.tool()
def my_tool(...):
    ...

License

MIT License