File size: 10,991 Bytes
f844f16 3708220 f844f16 3708220 f844f16 848db06 3708220 f844f16 848db06 f844f16 3708220 f844f16 3708220 f844f16 3708220 f844f16 3708220 f844f16 3708220 f844f16 3708220 848db06 3708220 848db06 f844f16 848db06 3708220 f844f16 848db06 f844f16 848db06 f844f16 3708220 f844f16 848db06 3708220 f844f16 3708220 f844f16 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 |
"""
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
}
) |