Real-time trading co-pilot with WebSocket streaming and live commentary for Binance cryptocurrency trading
Server has 15 tools with basic descriptions and input schemas present. However, critical gaps undermine quality: (1) Parameter descriptions are minimal or missing context (e.g., 'symbol' lacks format guidance like 'BTCUSDT'); (2) Output schemas are not documented, responses are inferred from code but not formally declared; (3) No error handling guidance, tools lack recovery hints for common failures (invalid symbol, insufficient balance, API errors); (4) Destructive tools (emergency_close_all, enter_long/short) lack confirmation patterns or dry-run support; (5) Tool descriptions are functional but lack WHEN/WHY context for LLM selection. Naming is clear and verb-based (get_, find_, enter_, close_), which is a strength. Average tool score: 52/100.
Close a specific position by ID.
EMERGENCY: Close ALL open positions immediately. Use this if something goes wrong.
Enter a LONG position (buy). I'll execute immediately and report back.
Enter a SHORT position (sell). I'll execute immediately and report back.
Find pairs currently breaking out (price near 24h high with momentum). Good for momentum/breakout trading.
Scan all Binance pairs and find the hottest trending ones. I'll narrate findings in real-time as I discover them.
Output schemas not documented. Tool descriptions state what is returned (e.g., 'Current price and symbol info') but do not formally declare the response structure. LLMs cannot plan downstream calls or extract fields reliably without documented schemas.
Destructive tools lack confirmation or dry-run patterns. emergency_close_all, enter_long, enter_short, and close_position execute immediately with no confirmation step. Agents can accidentally liquidate positions or lose capital without a safety gate.
Parameter descriptions lack format/constraint guidance. 'symbol' parameter appears in 8 tools but descriptions do not specify format (e.g., 'uppercase pair like BTCUSDT'), valid ranges, or examples. 'quantity' in enter_long/enter_short lacks min/max bounds.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | <=2025-11-25 | v2 |
Get 24-hour trading statistics for a symbol.
Get current account balances for all assets.
Get the current market price for a trading pair.
Get all currently open orders.
Get all currently open trading positions.
Get the order book (bids and asks) for a symbol.
Get trading statistics (wins, losses, total P&L, etc.).
Start monitoring and trading a symbol on the 1-minute chart. I'll constantly narrate what I'm seeing and doing. You can interrupt me anytime by pressing ESC and giving new instructions.
Stop the current trading session.
No error handling guidance. Tools do not document what happens on invalid symbol, insufficient balance, API rate limit, or network failure. LLMs receive raw errors with no recovery hints.
Tool descriptions lack WHEN/WHY context. Descriptions state WHAT the tool does but not when an LLM should choose it over similar tools or what prerequisites exist. E.g., find_hot_pairs vs find_breakouts, when should each be used?