Alpaca Trading
Trade stocks, options and crypto through an Alpaca brokerage account, and pull the market data behind the decision, in the same conversation.Read full description
alpacahqLABSCO SUMMARY
Real orders on a real account — the guardrail is you, not the server.
Positions and assets read back fine, and orders go through at the price and size you state. The failure mode the author calls out is name-to-code translation: asking for a stock by its Chinese name can resolve to the wrong ticker depending on the model, so naming the code with its .SH or .SZ suffix is the safe habit.
What it is
A server that connects to the Xuntou (迅投) QMT trading system through a local miniQMT install and exposes account and order operations over MCP.
What you get
- Account asset queries — what the account holds in cash and value
- Position queries — the current holdings
- Order placement, with the price and share count you name
- Order cancellation
- Ticker codes carry their exchange suffix, .SH or .SZ
Requirements
Python >= 3.10, uv, QMT trading permissions on your brokerage account, and miniQMT already running locally. On first launch the server asks for the miniQMT path and the fund account number and saves them to xttrader.yaml. Clients connect over SSE at http://localhost:8001/sse. The author states this is for study and exchange, and asks that you be careful pointing it at a live account.
Setup effort
Build from source — git clone https://github.com/nnquant/qmt-mcp-server
USE CASES
01
Checking what the account is holding right now
Ask for positions and account assets in the same thread you are doing research in.
02
Placing an order you have already decided on
State the price, the share count and the ticker code — the order goes to miniQMT directly.
03
Pulling an order back
Cancel a resting order without switching to the trading terminal.
04
Chaining screening and trading together
Combine it with another server that does the screening, and act on the shortlist in the same conversation.
BEFORE YOU INSTALL
What to know first
Requires
Python >= 3.10
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
This one is built from source rather than installed with a single command.
Server setup
TERMINAL
git clone https://github.com/nnquant/qmt-mcp-server
cd qmt-mcp-server
uv sync
uv run main.pyMCP Clients
Cursor
CURSOR · JSON
{
"mcpServers": {
"qmt-mcp-server": {
"url": "http://localhost:8001/sse"
}
}
}
