Jump to content

Connect Leeroopedia MCP: Equip your AI agents to search best practices, build plans, verify code, diagnose failures, and look up hyperparameter defaults.

Implementation:HKUDS AI Trader Get Agent System Prompt

From Leeroopedia


Knowledge Sources
Domains Prompt_Engineering, LLM_Agents
Last Updated 2026-02-09 14:00 GMT

Overview

Concrete tool for building context-rich system prompts with live market data injection for LLM trading agents.

Description

The get_agent_system_prompt function constructs the full system prompt for each trading day. It fetches yesterday's buy/sell prices via get_yesterday_open_and_close_price(), today's open prices via get_open_prices(), and current holdings via get_today_init_position(). These are formatted and injected into a template string that includes trading rules and the STOP_SIGNAL termination token.

The STOP_SIGNAL constant is defined as "<FINISH_SIGNAL>" and is injected into the prompt template so the agent knows how to signal completion.

Usage

Call this function at the start of each trading session (daily loop iteration) to generate the system prompt passed to create_agent(). Different prompt modules exist for each market: agent_prompt.py (US), agent_prompt_astock.py (A-share), agent_prompt_crypto.py (crypto).

Code Reference

Source Location

  • Repository: AI-Trader
  • File: prompts/agent_prompt.py
  • Lines: L62-88

Signature

STOP_SIGNAL = "<FINISH_SIGNAL>"  # L23

def get_agent_system_prompt(
    today_date: str,
    signature: str,
    market: str = "us",
    stock_symbols: Optional[List[str]] = None
) -> str:
    """
    Construct the system prompt for a trading agent on a given date.

    Args:
        today_date: Current trading date in YYYY-MM-DD format
        signature: Agent model name/identifier
        market: Market type ("us" or "cn")
        stock_symbols: Override symbol list (defaults to NASDAQ-100 or SSE-50)

    Returns:
        Formatted system prompt string with market data and trading rules
    """

Import

from prompts.agent_prompt import get_agent_system_prompt, STOP_SIGNAL

I/O Contract

Inputs

Name Type Required Description
today_date str Yes Trading date YYYY-MM-DD
signature str Yes Agent model name for identification
market str No Market type: "us" (default) or "cn"
stock_symbols Optional[List[str]] No Override symbol list (defaults to NASDAQ-100 or SSE-50)

Outputs

Name Type Description
system_prompt str Complete system prompt with date, positions, prices, rules, and STOP_SIGNAL

Usage Examples

Generate US Stock Prompt

from prompts.agent_prompt import get_agent_system_prompt

prompt = get_agent_system_prompt(
    today_date="2025-01-15",
    signature="gpt-4o",
    market="us"
)
# Returns prompt containing:
# - Date: 2025-01-15
# - Current positions from position.jsonl
# - Yesterday's close prices for all NASDAQ-100 stocks
# - Today's open prices
# - Trading rules and STOP_SIGNAL

Related Pages

Uses Heuristic

Implements Principle

Page Connections

Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment