Tools - Docs

Tools help agents to connect with external software components like API’s or other functions.

Note: You can use our pre-built tools or tools from other providers like Llama Hub. Optionally you can also use our base Tool class to create your own custom tool by providing the function, input and output Pydantic models.

Let’s look at an example to use prebuilt linkedin post tool. All prebuilt tools are available to export from lyzr_automata.tools.prebuilt_tools

from lyzr_automata.tools.prebuilt_tools import linkedin_image_text_post_tool

linkedin_post_tool = linkedin_image_text_post_tool(
    owner="urn:li:person:<YOUR_ID_HERE>",
    token="YOUR_TOKEN_HERE",
)

Tool class

Description

This section provides an overview of how to initialize an instance of a Tool object. A Tool is designed to encapsulate a specific operation or function, complete with metadata describing its purpose, and structured input and output specifications. This setup allows for standardized execution of tasks, ranging from simple operations like arithmetic calculations to more complex data processing or analysis functions. The use of Pydantic models for input and output ensures type safety and data validation, enhancing the robustness of the tool.

Parameter Table

name

str

The name of the tool, giving a clear indication of its functionality or the task it is designed to perform.

desc

str

A brief description of the tool’s purpose and how it operates, providing context to its users.

function

str

The specific function the tool executes. This should be a callable object that performs the tool’s main operation.

function_input

str

The Pydantic model that defines the structure, type, and validation rules for the input data. This ensures that the function receives data in the expected format.

function_output

str

The Pydantic model for the output data, specifying what the function returns. This model validates the output data structure and type, ensuring consistency and reliability in the tool’s output.

Implementation Notes

Pydantic Models: function_input and function_output are crucial for defining clear contracts for the tool’s operation. Pydantic models facilitate data validation and error handling, making the tool more reliable and easier to integrate.
Customization: The flexibility in defining function, function_input, and function_output allows for a wide range of tools to be created, from simple utilities to complex data processing pipelines.

Create your own tool with Tool base class

Create a function

 def multiply_numbers(a, b):
    """
    Multiply two numbers.

    Parameters:
    - a (int/float): The first number.
    - b (int/float): The second number.

    Returns:
    - int/float: The product of the two numbers.
    """
    return a * b

Create Pydantic input model

#INPUT
from pydantic import BaseModel

class MultiplyInput(BaseModel):
	      a: float  # Assuming we want to allow floating point numbers
	      b: float

Create Pydantic output model

#OUPUT
 class MultiplyOutput(BaseModel):
       result: float

Create a tool instance

from lyzr_automata import Tool

multiplication_tool = Tool(
        name="Multiplication tool",
        desc="multiplies two numbers",
        function=multiply_numbers,
        function_input=MultiplyInput,
        function_output=MultiplyOutput,
    )

Lyzr Automata