Skip to content
Nappai Documentation

Python Code Structured

The Python Code Structured component in Nappai enables users to execute structured Python code securely and in a controlled manner. It is designed to transform structured data class code into usable tools within a specific environment, applying security restrictions and custom configurations.

Relationship with Python

This component leverages Python to execute structured code safely. It ensures secure execution by implementing blacklists for functions and modules, and manages the creation and deletion of temporary files for code execution.

Inputs

  • Tool Code: Enter the data class code that defines the tool. This input supports real-time updates and uses a Python code schema.
  • Tool Name: Specify the name of the tool. This is a required field.
  • Description: Provide a brief description of the tool. This is a required field.
  • Return Directly: Indicate whether the tool should directly return the function’s output. The default value is False.
  • Tool Function: Select the function for additional expressions. This input supports real-time updates and includes a refresh button.
  • Parent Document Vectorstore: Specify the vectorstore of the parent document.
  • Classes: Advanced input for information about classes defined in the code.
  • Functions: Advanced input for information about functions defined in the code.

Outputs

The component produces two main outputs:

  • Tool: The result of the tool construction process.
  • Text: A text response generated by the component based on the input message.

These outputs can be used in workflows to automate tasks and processes within the Nappai system.

Usage Example

Imagine you want to automate a data analysis task. You can input your Python code that defines the analysis tool, specify its name and description, and configure whether it should return results directly. Once set up, the component will execute your code securely and provide the analysis results as output.

Templates

Currently, there are no specific templates where this component is used.

  • Sequential Task: A component that wraps around the CrewAI library for task management.
  • SQLAgent: Constructs an SQL agent from an LLM and tools.
  • OpenAI Tools Agent: Uses tools via OpenAI’s language models.

Tips and Best Practices

  • Always review your Python code for security vulnerabilities before inputting it into the component.
  • Use descriptive names and descriptions for your tools to make them easily identifiable.
  • Regularly update your tool functions to ensure they align with your current workflow needs.

Security Considerations

The component implements blacklists for Python functions and modules to ensure secure code execution. It also manages temporary files to prevent unauthorized access or data leaks.