CCXT MCP Server

What is CCXT MCP Server?

A high-performance cryptocurrency exchange integration server that uses the Model Context Protocol (MCP) and the CCXT library to connect large language models (LLMs) like Claude and GPT to 20+ cryptocurrency exchanges. It allows LLMs to access real-time market data and execute trading operations through a unified API.

How to use CCXT MCP Server?

Install globally via npm (npm install -g @mcpfun/mcp-server-ccxt) or run directly with npx @mcpfun/mcp-server-ccxt. Configure exchange API keys in a .env file. Start the server and add it to a MCP-compatible client (e.g., Claude Desktop) using mcp-server-ccxt as the command. Over 30 MCP tools are available for public market data and private account/trading operations.

Key features of CCXT MCP Server

• Supports 20+ cryptocurrency exchanges (spot, futures, swaps, etc.)
• Proxy configuration for accessing exchanges through proxies
• Optimized caching and adaptive rate limiting
• MCP standard compatible with LLMs like Claude and GPT
• LRU caching with configurable TTLs per data type
• Efficient exchange connection management with error handling

Use cases of CCXT MCP Server

• Query current prices, order books, and OHLCV data across multiple exchanges
• Compare cryptocurrency prices and market depth across different exchanges
• Retrieve account balances and open positions (with API keys)
• Place market or limit orders, cancel orders, and manage futures positions
• Monitor funding rates, leverage tiers, and exchange status in real time

FAQ from CCXT MCP Server

What runtime/dependencies does it require?

Node.js and npm. The server is published as an npm package and depends on the CCXT library (version 4.0.0) which is MIT licensed.

How are API keys handled?

API keys and secrets are stored in a .env file (never committed to version control) and loaded as environment variables. Recommended security practices include using dedicated keys, limiting permissions, and enabling IP whitelisting.

Where does market data come from and how fresh is it?

All data is fetched live from the connected exchange via their public APIs. Caching is applied with different TTLs: ticker data (10 seconds), order book (5 seconds), and market data (1 hour).

What are the rate limiting and request limits?

The server implements adaptive rate limiting that automatically adjusts request rates based on exchange responses and uses exponential backoff on errors. Actual limits depend on each exchange's own API rate limits.

How does the server communicate with LLMs?

It uses the Model Context Protocol (MCP) over standard input/output (stdin/stdout), making it compatible with any MCP-compatible client such as Claude Desktop.