QBT Labs uses OpenMM to run live market-making workflows across centralized exchanges while exposing the same execution layer to AI agents through SDKs and MCP tools. It is the production proof point behind our agent-payment infrastructure.
4
CEX venues: MEXC, Gate.io, Bitget, Kraken
13+
MCP trading tools for agent clients
MIT
Open-source SDK and MCP tooling
24/7
Always-on liquidity and monitoring
Market making is API-heavy, capital-sensitive, and unforgiving. A production desk needs exchange connectivity, inventory policy, risk controls, order placement, fill monitoring, reporting, and failover. Those are the same primitives autonomous agents need when they pay for tools, execute trades, and stay inside budget.
OpenMM provides a TypeScript SDK and MCP server for balances, order books, order placement, strategy execution, and portfolio visibility. QBT Labs uses it as live infrastructure, not a demo: the stack connects to MEXC, Gate.io, Bitget, and Kraken and is designed to expose trading workflows to Claude, Cursor, Windsurf, and other MCP clients.
An agent that trades also needs to pay: market data, paid APIs, execution services, settlement, and receipts. OpenMM is the workload that forces the QBT stack to handle real budgets, signer isolation, audit logs, and machine-readable payment flows.
Use OpenMM for trading execution, x402 for agent-paid APIs, and QBT Labs' policy layer for controlled autonomous spend.