Adding Pyton REPL to Tool

In this guide, we'll extend a LangChain-based agent with Python REPL functionality. By combining the Tavily Search Results tool for information lookup with a Python REPL tool for on-the-fly calculations, your agent can handle both data retrieval and computation seamlessly.

Prerequisites

• Tavily API key stored in the TAVILY_API_KEY environment variable
• Access to OpenAI's Chat API via langchain-openai

1. Import Modules and Initialize Tools

from langchain_core.prompts.chat import ChatPromptTemplate, MessagesPlaceholder
from langchain_community.chat_message_histories import ChatMessageHistory
from langchain_core.runnables.history import RunnableWithMessageHistory

from langchain_openai import ChatOpenAI
from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_experimental.tools import PythonREPLTool

from langchain.agents import create_tool_calling_agent
from langchain.agents import AgentExecutor

import os

# Set your Tavily API key
TAVILY_API_KEY = os.getenv("TAVILY_API_KEY")

search = TavilySearchResults()
python_repl = PythonREPLTool()

2. Build the Chat Prompt Template

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a helpful assistant. Think step by step before responding."),
    ("placeholder", "{chat_history}"),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}")
])

3. Initialize the Agent Executor

llm = ChatOpenAI()
tools = [search, python_repl]
message_history = ChatMessageHistory()

agent = create_tool_calling_agent(llm, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

4. Example Queries

Below are common examples demonstrating search and computation capabilities.

1. Tournament Schedule

   response = agent_executor.invoke({"input": "When is the ICC Men's T20 World Cup scheduled?"})
   print(response["output"])
   

2. Hosting Countries

   response = agent_executor.invoke(
       {"input": "Which countries are hosting?"}, 
       config={"configurable": {"$session_id": "session"}}
   )
   print(response["output"])
   

3. Date Calculation

   response = agent_executor.invoke({
       "input": "Today's date is May 1st, 2024. How many days before the first match starts?"
   })
   print(response["output"])
   

Under the Hood: Python REPL Execution

# Code executed in the Python REPL environment
from datetime import datetime

start_date = datetime(2024, 6, 1)
today_date = datetime(2024, 5, 1)
days_until_start = (start_date - today_date).days
print(days_until_start)

This code prints:

31

And the agent returns:

There are 31 days left before the first match of the ICC Men's T20 World Cup 2024 starts on June 1, 2024.

By combining the Tavily search tool for tournament data and the Python REPL tool for on-the-fly math, the agent can answer complex queries accurately and interactively.

Links and References

• LangChain Documentation
• OpenAI Chat API
• Tavily Search Results Tool

Stay tuned for more demos and advanced capabilities in upcoming articles!