AI Zhangle A-Share Paper Trading

Problem

A common pain point in A-share paper trading is that users may mention a ticker by name, such as “Moutai,” instead of a code, while 000001 can mean different instruments on different exchanges. Before ordering, the caller must verify price, available funds, and sellable quantity. After trading, it must distinguish pending orders, partial fills, and historical trades. Mapping natural language directly to trading APIs can easily miss parameters or misclassify errors.

How It Works

The skill wraps the Huatai Securities AI Zhangle backend CLI: searchStock, getQuote, getAccountBalance, getPositions, submitOrder, cancelOrder, listPendingOrders, and listTradeHistory.
- Resolve tickers: use searchStock first to get the stockCode + exchange pair.
- Check state: use getQuote for price, limit-up/limit-down, and suspension; use account and position endpoints for funds and holdings.
- Execute trading: use submitOrder to place orders and cancelOrder to cancel, then inspect listPendingOrders and listTradeHistory for pending or filled activity.
- Handle errors: branch on error.category, such as auth, validation, business, and network.

Boundaries

It supports A-share paper trading only, not HK, US, futures, FX, or real capital trading. Amounts are in CNY, percentages are percent values, sellable quantity should use availableQuantity, and the integration requires HT_APIKEY and PAPER_TRADING_API_URL.