OKX MCP Server

This project creates a Model Context Protocol (MCP) server that fetches real-time cryptocurrency data from the OKX exchange. It allows AI assistants like Claude to access up-to-date cryptocurrency price information and historical data through defined tools with enhanced visualization capabilities and WebSocket live updates.

Features

get_price: Fetches the latest price data for a cryptocurrency trading pair with visual formatting
get_candlesticks: Retrieves historical candlestick data with visualization options including ASCII charts
subscribe_ticker: Subscribes to real-time WebSocket updates for a trading pair
get_live_ticker: Retrieves the latest live data from WebSocket connection
unsubscribe_ticker: Stops receiving updates for a specific trading pair

Prerequisites

Node.js (v16 or higher recommended)
npm or yarn
VSCode with Claude extension (if using VSCode integration)
Claude Desktop (if not using VSCode)

Installation

Clone the repository

git clone https://github.com/yourusername/okx-mcp-server.git
cd okx-mcp-server

Install dependencies
```
npm install
```

Add the WebSocket dependency

npm install ws
npm install --save-dev @types/ws

Build the project
```
npm run build
```
Make the compiled script executable
```
chmod +x build/index.js
```

Usage

Running the Server Directly

You can run the server directly with:

npm start

Or:

node build/index.js

Testing with MCP Inspector

To test your MCP server before integration:

npx @modelcontextprotocol/inspector node build/index.js

In the inspector, you can test:

get_price with input: { "instrument": "BTC-USDT", "format": "markdown" }
get_candlesticks with input: { "instrument": "BTC-USDT", "bar": "1m", "limit": 10, "format": "markdown" }
subscribe_ticker with input: { "instrument": "BTC-USDT" }
get_live_ticker with input: { "instrument": "BTC-USDT", "format": "markdown" }
unsubscribe_ticker with input: { "instrument": "BTC-USDT" }

Each tool supports different visualization formats:

format: Choose between markdown, json, or table (for candlesticks)

Integration with VSCode

Install the Claude extension for VSCode

Configure the MCP server in VSCode settings:

Create or edit the following file:

~/Library/Application Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json

Add the following configuration:

{
  "mcpServers": {
    "okx": {
      "command": "node",
      "args": ["/absolute/path/to/okx-mcp-server/build/index.js"],
      "disabled": false,
      "autoApprove": []
    }
  }
}

Replace /absolute/path/to/okx-mcp-server/build/index.js with your actual file path

Restart VSCode or reload the Claude extension

Integration with Claude Desktop

If using Claude Desktop, check their documentation for the appropriate location to place MCP configuration settings.

Example Prompts

Once integrated, you can ask Claude:

"What's the current price of Bitcoin (BTC-USDT)? Show me a nicely formatted display."
"Show me the price trend of Ethereum (ETH-USDT) over the last hour using 5-minute intervals with a visual representation."
"Compare the current prices of BTC-USDT and ETH-USDT."
"Analyze the most recent 20 candlesticks for SOL-USDT with 1-minute intervals and display them in a table format."
"Is the current price of BTC-USDT higher or lower than its 24-hour high? Visualize the price range."
"Subscribe to live updates for Bitcoin price and tell me when it changes."
"Show me the latest real-time data for Ethereum."
"Monitor BTC-USDT in real-time and alert me if the price moves by more than 1% in either direction."

Data Visualization Features

The enhanced MCP server provides rich data visualization:

Price Data Visualization

Formatted markdown output with clear price information
Visual price range bar showing where current price sits between 24h high/low
Directional indicators (▲/▼) for price changes
Formatted numbers for better readability

Candlestick Data Visualization

Markdown format: Includes summary statistics, ASCII chart of price movements, and recent price action table
Table format: Clean tabular presentation of candlestick data
JSON format: Raw data for programmatic use

Format Options

get_price: Supports markdown (default) or json formats
get_candlesticks: Supports markdown (default), table, or json formats
get_live_ticker: Supports markdown (default) or json formats

WebSocket Real-time Updates

This server includes WebSocket support for receiving live data from OKX:

How WebSockets Work

Establishes a persistent connection to OKX's WebSocket API
Receives push updates whenever prices change
Maintains a cache of the latest data for each subscribed instrument
Automatically reconnects if the connection is lost

WebSocket Tools

subscribe_ticker: Starts a subscription for real-time ticker updates
get_live_ticker: Gets the latest data from active subscriptions
unsubscribe_ticker: Ends a subscription when monitoring is no longer needed

Use Cases for Real-time Updates

Monitoring price movements in real-time
Setting up price alerts
Tracking market volatility as it happens
Making informed trading decisions with the latest data

Environment Variables (Optional)

This basic implementation uses OKX's public API endpoints that don't require authentication. If you extend the server to use authenticated endpoints, you may want to add these environment variables:

OKX_API_KEY=your_api_key
OKX_API_SECRET=your_api_secret
OKX_API_PASSPHRASE=your_api_passphrase

Security Notes

The current implementation only uses public OKX API endpoints, so no API keys are required
No sensitive data is stored in the codebase
It's safe to commit this code to Git as is

Extending the Server

You can extend this MCP server by:

Adding more tools for other OKX API endpoints
Implementing authenticated endpoints with API keys
Adding support for other exchanges
Further enhancing the visualizations with more complex charts
Adding technical analysis indicators (RSI, MACD, Moving Averages)
Implementing price comparison tools for multiple cryptocurrencies
Adding market summary features for broader market analysis

License

MIT