from autogen import AssistantAgent from tools.knowledge_base_tool import search_similar_solution from utility.llm_config import llm_config def get_knowledge_base_agent(): knowledge_agent = AssistantAgent( name="KnowledgeBaseAgent", system_message=( "You are an IT support assistant that retrieves solutions to user issues. " "Always call the 'search_similar_solution' tool with the query to get a matching solution. " "After calling, summarize the top solution and respond with TERMINATE." ), llm_config=llm_config, # Required for tool registration code_execution_config={"use_docker": False}, ) # Register tool with LLM and executor # 1. LLm knows when to call the tool knowledge_agent.register_for_llm( name="search_similar_solution", description="Searches for top IT solutions from a knowledge base using a vector similarity search. Accepts query and top_k." )(search_similar_solution) # 2. Executed the tool knowledge_agent.register_for_execution( name="search_similar_solution" )(search_similar_solution) return knowledge_agent