Creating tools for agents in LangChain - Try It Yourself

Creating Tools for Agents with LangChain

📖 Scenario: You are building a simple assistant that can perform specific tasks using tools. These tools help the assistant answer questions or perform actions by connecting to external functions.

🎯 Goal: Create a LangChain agent that uses a custom tool to answer questions about the current date and time.

📋 What You'll Learn

Create a tool function that returns the current date and time as a string.

Create a LangChain Tool object using the tool function.

Create an agent that uses the tool to answer questions.

Run the agent with a sample question about the current date and time.

💡 Why This Matters

🌍 Real World

Agents with tools help automate tasks by connecting language models to real functions, making assistants smarter and more useful.

💼 Career

Understanding how to create and use tools in LangChain is valuable for building AI assistants, chatbots, and automation systems in software development roles.

1

Create a tool function to get current date and time

Write a Python function called get_current_datetime that returns the current date and time as a string using datetime.now() and strftime with format "%Y-%m-%d %H:%M:%S".

LangChain

# Your code here
def get_current_datetime():
    pass

Need a hint?

Use from datetime import datetime and return datetime.now().strftime("%Y-%m-%d %H:%M:%S").

2

Create a LangChain Tool using the function

Import Tool from langchain.agents and create a variable called datetime_tool that is a Tool object. Set its name to "CurrentDateTime", func to get_current_datetime, and description to "Returns the current date and time as a string.".

LangChain

from datetime import datetime

def get_current_datetime():
    return datetime.now().strftime("%Y-%m-%d %H:%M:%S")

# Your code here

Need a hint?

Use Tool(name=..., func=..., description=...) to create the tool.

3

Create an agent using the tool

Import initialize_agent and AgentType from langchain.agents. Create a variable called agent by calling initialize_agent with a list containing datetime_tool, an llm set to None (placeholder), and agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION. This sets up the agent to use the tool.

LangChain

from datetime import datetime
from langchain.agents import Tool, initialize_agent, AgentType

def get_current_datetime():
    return datetime.now().strftime("%Y-%m-%d %H:%M:%S")

datetime_tool = Tool(
    name="CurrentDateTime",
    func=get_current_datetime,
    description="Returns the current date and time as a string."
)

# Your code here

Need a hint?

Use initialize_agent(tools=[datetime_tool], llm=None, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION).

4

Run the agent with a sample question

Call agent.run with the string "What is the current date and time?" to get the answer using the tool.

LangChain

from datetime import datetime
from langchain.agents import Tool, initialize_agent, AgentType

def get_current_datetime():
    return datetime.now().strftime("%Y-%m-%d %H:%M:%S")

datetime_tool = Tool(
    name="CurrentDateTime",
    func=get_current_datetime,
    description="Returns the current date and time as a string."
)

agent = initialize_agent(
    tools=[datetime_tool],
    llm=None,
    agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION
)

# Your code here

Need a hint?

Use agent.run("What is the current date and time?") to ask the agent.