""" Research Agent - Information gathering and research tasks The Research Agent is responsible for: 1. Gathering information from multiple sources (web, Wikipedia, arXiv) 2. Searching for relevant context and facts 3. Compiling research results in a structured format 4. Returning citations and source information """ import os from typing import Dict, Any, List from langchain_core.messages import BaseMessage, HumanMessage, SystemMessage, AIMessage from langgraph.types import Command from langchain_groq import ChatGroq from langchain_core.tools import BaseTool from observability import agent_span, tool_span from dotenv import load_dotenv # Import LangChain-compatible tools from langgraph_tools import get_research_tools load_dotenv("env.local") def load_research_prompt() -> str: """Load the research-specific prompt""" try: with open("archive/prompts/retrieval_prompt.txt", "r") as f: return f.read() except FileNotFoundError: return """ You are a research specialist focused on gathering accurate information. Your goals: 1. Search for factual, current, and relevant information 2. Use multiple sources to verify facts 3. Provide clear citations and sources 4. Structure findings in an organized manner When researching: - Use web search for current information and facts - Use Wikipedia for encyclopedic knowledge - Use ArXiv for academic and technical topics - Cross-reference information across sources - Note any conflicting information found Format your response as: ### Research Strategy [Describe what searches are needed] ### Findings [Key information discovered] ### Key Facts - Fact 1 - Fact 2 - Fact 3 ### Sources - Source 1 - Source 2 """ def research_agent(state: Dict[str, Any]) -> Command: """ Research Agent node that gathers information using LangChain tools. Returns Command with research results appended to research_notes. """ print("🔍 Research Agent: Gathering information...") try: # Get research prompt research_prompt = load_research_prompt() # Initialize LLM with tool binding llm = ChatGroq( model="llama-3.3-70b-versatile", temperature=0.3, # Slightly higher for research creativity max_tokens=2048 ) # Get LangChain-compatible research tools research_tools = get_research_tools() # Bind tools to LLM for function calling llm_with_tools = llm.bind_tools(research_tools) # Create agent span for tracing with agent_span( "research", metadata={ "tools_available": len(research_tools), "user_id": state.get("user_id", "unknown"), "session_id": state.get("session_id", "unknown") } ) as span: # Extract user query messages = state.get("messages", []) user_query = "" for msg in messages: if isinstance(msg, HumanMessage): user_query = msg.content break # Build research request research_request = f""" You must research the following question using the available tools. Do not answer from memory alone. Question: {user_query} Current research status: {len(state.get('research_notes', ''))} characters already gathered CRITICAL: You MUST use the available research tools to gather information. Do not provide an answer without using tools. Available tools: - tavily_search_results_json: For current web information - wikipedia_search: For encyclopedic knowledge - arxiv_search: For academic papers Instructions: 1. ALWAYS use tavily_search_results_json for current information 2. Use wikipedia_search for general knowledge topics 3. Use arxiv_search for academic/technical topics if relevant 4. You must call at least one tool - preferably multiple tools 5. Analyze and synthesize the information from the tools 6. Provide structured findings with sources Start by calling the appropriate research tools to gather information about this question. """ # Create messages for research research_messages = [ SystemMessage(content=research_prompt), HumanMessage(content=research_request) ] # Get research response with tool calls response = llm_with_tools.invoke(research_messages) # Debug: Check if tools were called print(f"🔍 Research response type: {type(response)}") print(f"🔍 Has tool_calls attribute: {hasattr(response, 'tool_calls')}") if hasattr(response, 'tool_calls'): print(f"🔍 Tool calls: {response.tool_calls}") else: print(f"🔍 Response content preview: {str(response)[:200]}...") # Process tool calls if any tool_results = [] if hasattr(response, 'tool_calls') and response.tool_calls: print(f"🛠️ Executing {len(response.tool_calls)} research tools") for tool_call in response.tool_calls: try: # Find the tool by name tool = next((t for t in research_tools if t.name == tool_call['name']), None) if tool: # Execute the tool with tool_span(tool.name, metadata={"args": tool_call.get('args', {})}) as tool_span_ctx: result = tool.invoke(tool_call.get('args', {})) tool_results.append(f"**{tool.name}**: {result}") if tool_span_ctx: tool_span_ctx.update_trace(output={"result": str(result)[:200] + "..."}) else: tool_results.append(f"**{tool_call['name']}**: Tool not found") except Exception as e: print(f"⚠️ Tool {tool_call.get('name', 'unknown')} failed: {e}") tool_results.append(f"**{tool_call.get('name', 'unknown')}**: Error - {str(e)}") else: print("⚠️ No tool calls detected - LLM did not choose to use any tools") # Force tool usage for research questions print("🔧 Forcing tool usage for research...") # Manually call appropriate tools based on query forced_tools = [] # Always try web search for current info tavily_tool = next((t for t in research_tools if t.name == "tavily_search_results_json"), None) if tavily_tool: try: print("🌐 Forcing Tavily web search...") result = tavily_tool.invoke({"query": user_query}) forced_tools.append(f"**tavily_search_results_json (forced)**: {result}") except Exception as e: print(f"⚠️ Forced Tavily search failed: {e}") forced_tools.append(f"**tavily_search_results_json (forced)**: Error - {str(e)}") # Try Wikipedia for general knowledge wiki_tool = next((t for t in research_tools if t.name == "wikipedia_search"), None) if wiki_tool: try: print("📚 Forcing Wikipedia search...") result = wiki_tool.invoke({"query": user_query}) forced_tools.append(f"**wikipedia_search (forced)**: {result}") except Exception as e: print(f"⚠️ Forced Wikipedia search failed: {e}") forced_tools.append(f"**wikipedia_search (forced)**: Error - {str(e)}") tool_results = forced_tools # Compile research findings if tool_results: research_findings = "\n\n".join(tool_results) # Ask LLM to analyze the tool results analysis_request = f""" Based on the research results below, provide a structured analysis: Original Question: {user_query} Research Results: {research_findings} Current research status: {len(state.get('research_notes', ''))} characters already gathered Instructions: 1. Analyze the search results for relevant information 2. Extract key facts that help answer the question 3. Note any important details or findings 4. Identify if additional specific searches might be needed 5. Structure your findings clearly with citations Please provide a comprehensive analysis of the research findings. """ analysis_messages = [ SystemMessage(content=research_prompt), HumanMessage(content=analysis_request) ] analysis_response = llm.invoke(analysis_messages) analysis_content = analysis_response.content if hasattr(analysis_response, 'content') else str(analysis_response) # Format final research results formatted_results = f""" ### Research Iteration {state.get('loop_counter', 0) + 1} {analysis_content} ### Raw Tool Results {research_findings} --- """ else: # No tools were called, use the LLM response directly response_content = response.content if hasattr(response, 'content') else str(response) formatted_results = f""" ### Research Iteration {state.get('loop_counter', 0) + 1} {response_content} --- """ print(f"📝 Research Agent: Gathered {len(formatted_results)} characters") # Update trace if span: span.update_trace(output={ "research_length": len(formatted_results), "tools_used": len(tool_results), "findings_preview": formatted_results[:300] + "..." }) # Return command to proceed back to lead agent for decision return Command( goto="lead", update={ "research_notes": formatted_results } ) except Exception as e: print(f"❌ Research Agent Error: {e}") # Return with error information error_result = f""" ### Research Error An error occurred during research: {str(e)} """ return Command( goto="lead", update={ "research_notes": error_result } )