Implementation:HKUDS AI Trader BaseAgent Initialize

Knowledge Sources	AI-Trader LangChain Agents
Domains	LLM_Agents, Infrastructure
Last Updated	2026-02-09 14:00 GMT

Overview

Concrete tool for initializing the BaseAgent's MCP client, tool set, and LLM model instance for trading sessions.

Description

The BaseAgent.initialize() async method creates the MultiServerMCPClient from self.mcp_config, fetches tools via await self.client.get_tools(), and instantiates ChatOpenAI (or DeepSeekChatOpenAI for deepseek models) with the configured API credentials. The method stores results on self.client, self.tools, and self.model for use in subsequent trading sessions.

Usage

Call await agent.initialize() after constructing the BaseAgent instance and before calling agent.run_date_range(). Must be awaited as it performs async MCP connections.

Code Reference

Source Location

• Repository: AI-Trader
• File: agent/base_agent/base_agent.py
• Lines: L330-404

Signature

async def initialize(self) -> None:
    """
    Initialize MCP client, tools, and LLM model.

    Sets:
        self.client: MultiServerMCPClient connected to MCP servers
        self.tools: List of available MCP tools
        self.model: ChatOpenAI or DeepSeekChatOpenAI instance

    Returns:
        None
    """

Import

from agent.base_agent.base_agent import BaseAgent
# Then: await agent.initialize()

I/O Contract

Inputs

Name	Type	Required	Description
self.mcp_config	Dict[str, Dict]	Yes	MCP server configuration (URLs and transport types)
self.basemodel	str	Yes	LLM model name (e.g., "gpt-4o", "deepseek-chat")
self.openai_api_key	str	Yes	API key for the LLM provider
self.openai_base_url	str	Yes	Base URL for the LLM API endpoint

Outputs

Name	Type	Description
self.client	MultiServerMCPClient	Connected MCP client
self.tools	List	Available MCP tools from all connected servers
self.model	ChatOpenAI	LLM model instance ready for inference

Usage Examples

Initialize Agent

from agent.base_agent.base_agent import BaseAgent

agent = BaseAgent(
    signature="gpt-4o",
    basemodel="gpt-4o",
    openai_api_key="sk-...",
    openai_base_url="https://api.openai.com/v1"
)

await agent.initialize()
# agent.client, agent.tools, agent.model are now set

Related Pages

Requires Environment

• Environment:HKUDS_AI_Trader_Python_LangChain_Runtime
• Environment:HKUDS_AI_Trader_API_Credentials
• Environment:HKUDS_AI_Trader_MCP_Services

Uses Heuristic

• Heuristic:HKUDS_AI_Trader_DeepSeek_Tool_Args_Workaround

Implements Principle

• Principle:HKUDS_AI_Trader_LangChain_Agent_Initialization