47 tools covering the full LMEX REST API — spot and futures market data, order placement, wallet management, and position control — usable from any MCP-compatible AI agent. Testnet mode included.
Market data, order placement, position management, leverage and risk limits, settlement currency — every REST endpoint available as a typed MCP tool.
Set LMEX_TESTNET=true to point at test.lmex.io. Test order flows, position management, and wallet transfers without touching real funds.
LIMIT, MARKET, STOP, TRIGGER, OCO (One-Cancels-Other), PEG (algo index-tracking), and reduce-only futures orders. Full time-in-force support: GTC, IOC, FOK, HALFMIN through MONTH.
lmex_spot_cancel_all_after and lmex_futures_cancel_all_after cancel all open orders after a TTL. Set timeout to 0 to disable. Safety net for unattended agents.
lmex_futures_bind_tpsl attaches take-profit and stop-loss levels to an open position in a single call.
All authenticated requests are signed automatically: HMAC-SHA384(secret, path + nonce + body). You supply the key and secret, the server handles the signature.
npx @lmex-official/lmex-mcp-servernpx @lmex-official/lmex-mcp-serverRequires Node.js 18+. Set LMEX_API_KEY, LMEX_API_SECRET, and LMEX_TESTNET in your MCP client config. Public market data endpoints require no key.
This server can place real orders on a live exchange. Set LMEX_TESTNET=true in your MCP config to point at the testnet environment until you've verified your agent's behaviour.
Public market data endpoints (market summary, OHLCV, orderbook, trades) require no API key. Only trading and wallet endpoints need credentials — use minimum required permissions.
Public endpoints need no key. Spot/futures trading needs the Trading permission. History and fee endpoints need Read. Wallet transfers need Transfer. Request only what you need.
Set LMEX_TESTNET=true in your MCP client config env block to use test.lmex.io. Set it to false (or omit it) for production at api.lmex.io. Create separate testnet API keys at test.lmex.io.
Query endpoints: 15 req/s per API, 30 req/s per user. Order endpoints: 75 req/s per API and per user. The server surfaces 429 errors — add delays for bulk operations.