Spaces:
Runtime error
Runtime error
| """ | |
| Setup for the legal assistant agent. | |
| """ | |
| from langchain.agents import AgentExecutor, create_tool_calling_agent | |
| from langchain.prompts import ChatPromptTemplate | |
| from AI_core.config import AGENT_LLM | |
| from AI_core.tools import tools | |
| # Create the main legal assistant agent | |
| system_prompt = """ | |
| You are a helpful assistant that can use multiple tools to answer questions. | |
| You have access to multiple specialized tools: | |
| 1. Document Summarization - For summarizing legal documents | |
| 2. Case Report Generation - For creating comprehensive legal case reports | |
| 3. Evidence Analysis - For analyzing legal evidence using advanced research capabilities | |
| (Use this tool when you need updated information regarding laws and public information) | |
| 4. Legal Q&A - For answering legal questions using a knowledge base | |
| 5. Legal Element Extraction - For extracting specific elements from legal texts | |
| """ | |
| agent_prompt = ChatPromptTemplate.from_messages( | |
| [ | |
| ("system", system_prompt), | |
| ("human", "{input}"), | |
| ("placeholder", "{agent_scratchpad}"), | |
| ] | |
| ) | |
| agent = create_tool_calling_agent(AGENT_LLM, tools, agent_prompt) | |
| # Create agent executor | |
| agent_executor = AgentExecutor( | |
| agent=agent, | |
| tools=tools, | |
| verbose=True, | |
| handle_parsing_errors=True, | |
| max_iterations=5 | |
| ) |