Docs
How To Guides
Create a New Tool

Create a New Tool

This guide will walk you through creating a new FunctionTool to compute the nth Fibonacci number.

Import the Required Modules

Begin by importing the necessary modules to create the tool. Create a new Python file titled main.py and add the following import statements:

main.py
from npiai import FunctionTool, function

Define the Tool Class

Next, define the MyTool class that inherits from FunctionTool. In the class constructor, specify the tool's name and description.

main.py
from npiai import FunctionTool, function
 
class MyTool(FunctionTool):
    def __init__(self):
        super().__init__(
            name='my_tool',
            description='test tool',
        )

Implement the Fibonacci Function

Add the fibonacci function to the MyTool class. This function calculates the nth Fibonacci number recursively.

main.py
from npiai import FunctionTool, function
 
class MyTool(FunctionTool):
    def __init__(self):
        super().__init__(
            name='my_tool',
            description='test tool',
        )
 
    def fibonacci(self, n):
        if n == 0:
            return 0
        if n == 1:
            return 1
        return self.fibonacci(n - 1) + self.fibonacci(n - 2)

Add Type Hints and Docstrings

Type hints and docstrings are essential for NPi to generate function schemas. Add type hints to the fibonacci function's parameters, and include a docstring describing the function's purpose and arguments:

main.py
from npiai import FunctionTool, function
 
class MyTool(FunctionTool):
    def __init__(self):
        super().__init__(
            name='my_tool',
            description='test tool',
        )
 
    def fibonacci(self, n: int) -> int:
        """
        Get the nth Fibonacci number.
 
        Args:
            n: The index of the Fibonacci number in the sequence.
        """
        if n == 0:
            return 0
        if n == 1:
            return 1
        return self.fibonacci(n - 1) + self.fibonacci(n - 2)

Annotated the Function with @function

Annotate the fibonacci function with the @function decorator to indicate that it is a LLM-callable function within the tool.

main.py
from npiai import FunctionTool, function
 
class MyTool(FunctionTool):
    def __init__(self):
        super().__init__(
            name='my_tool',
            description='test tool',
        )
 
    @function
    def fibonacci(self, n: int) -> int:
        """
        Get the nth Fibonacci number.
 
        Args:
            n: The index of the Fibonacci number in the sequence.
        """
        if n == 0:
            return 0
        if n == 1:
            return 1
        return self.fibonacci(n - 1) + self.fibonacci(n - 2)

Run the Tool

Finally, we can now run the tool in function mode or agent mode to interact with it. Refer to the Function Mode and Agent Mode section for detailed instructions.

async def main():
    async with MyTool() as my_tool:
        # do something with my_tool.tools
        print(my_tool.tools)
 
if __name__ == '__main__':
    import asyncio
    asyncio.run(main())

Note that the above example uses the context manager (opens in a new tab) to automatically starts and ends the tool. You can also manually start and stop the tool using the start() and stop() methods:

async def main():
    my_tool = MyTool()
    await my_tool.start()
    # do something with my_tool.tools
    print(my_tool.tools)
    await my_tool.end()
 
if __name__ == '__main__':
    import asyncio
    asyncio.run(main())
Schema GenerationUse Official Tool