OpenMM MCP Server — AI-native crypto trading infrastructure
Give AI agents structured access to live crypto markets: tickers, order books, balances, order execution, grid strategies, and Cardano DEX discovery through the Model Context Protocol.
Quick start
Install the server, add exchange API keys, then connect it to any MCP-compatible AI client.
# Run OpenMM MCP with npx
npx @qbtlabs/openmm-mcp
# Or install as a project dependency
npm install @qbtlabs/openmm-mcp
# Docs
open https://docs.openmm.io/quickstart13 trading tools exposed through MCP
Use cases
Live API reference and playground
The OpenMM API reference is hosted separately so developers can inspect endpoints, SDK types, and integration examples without waiting for a local setup.
Integration examples
Claude Desktop / Claude Code
Add OpenMM MCP as a stdio server and ask Claude to inspect balances, summarize order books, or prepare a dry-run grid strategy before execution.
npx @qbtlabs/openmm-mcpCursor / Windsurf
Use the MCP server as a trading-infrastructure tool layer while coding strategies, dashboards, or portfolio automations.
Frequently asked questions
What is the OpenMM MCP server?
OpenMM MCP is QBT Labs' open-source Model Context Protocol server for crypto trading. It exposes market data, balances, orders, grid strategies, and Cardano DEX discovery tools to AI clients such as Claude Desktop, Cursor, Windsurf, and other MCP-compatible runtimes.
Which exchanges does it support?
OpenMM MCP is built on the OpenMM execution stack and supports MEXC, Gate.io, Bitget, Kraken, and Cardano DEX discovery today. The architecture is adapter-based so additional venues can be added without changing the agent interface.
Is the MCP server free to use?
Yes. OpenMM MCP is open source and MIT licensed. Teams can self-host it, inspect the code, and connect it to their own exchange API keys. QBT Labs offers optional managed services for production trading operations.
Can an AI agent place real orders through it?
Yes, if you configure trading API keys with order permissions. For safety, teams should start with read-only keys or dry-run strategy previews, then add order permissions with exchange-level restrictions, withdrawal disabled, and external risk controls.