Voitta

A Python framework for routing, automating, and orchestrating LLM tool calls. Voitta simplifies the integration of AI agents with external tools and APIs, enabling more powerful and flexible AI applications.

Features

• Tool Call Routing: Seamlessly route LLM tool calls to the appropriate handlers
• Flexible Configuration: Define your tools and routing logic using YAML or Python
• Framework Agnostic: Works with any LLM provider or framework
• Extensible Architecture: Easily add custom tools and integrations
• Observability: Monitor and debug tool calls with built-in logging

Installation

From PyPI (Stable Release)

pip install voitta

From TestPyPI (Pre-release Versions)

To install the latest pre-release version from TestPyPI:

pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple/ voitta

The --extra-index-url flag is needed to fetch dependencies from the main PyPI repository, as TestPyPI may not have all the required dependencies.

You can also specify a particular version:

pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple/ voitta==0.2.3

Note: Pre-release versions on TestPyPI may contain experimental features and bugs. Use in production environments at your own risk.

Quick Start

from voitta import Voitta

# Initialize Voitta
voitta = Voitta()

# Register a tool handler
@voitta.tool("get_weather")
def get_weather(location, unit="celsius"):
    # Implementation to fetch weather data
    return {"temperature": 22, "condition": "sunny", "location": location, "unit": unit}

# Process an LLM tool call
result = voitta.process_tool_call({
    "name": "get_weather",
    "arguments": {"location": "San Francisco", "unit": "fahrenheit"}
})

print(result)  # Output: {"temperature": 72, "condition": "sunny", "location": "San Francisco", "unit": "fahrenheit"}

Usage

For detailed usage examples and documentation, please refer to:

• Voitta Example Repository
• Voitta Official Website
• Voitta on PyPI

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.