Skip to content

Installation

Get started with OpenTrace in just a few steps. Choose the installation method that works best for your needs.

Quick Installation

Recommended

Install OpenTrace from PyPI for the latest stable release: 

pip install trace-opt

Development Installation

For Contributors

Clone the repository and install in editable mode to contribute or access the latest features: 

git clone https://github.com/AgentOpt/OpenTrace.git
cd OpenTrace
pip install -e .

Requirements

System Requirements

  • Python 3.8+ - OpenTrace requires Python 3.8 or higher
  • No core dependencies - Basic tracing functionality works out of the box
  • Optional: LiteLLM - Required only if using opto.optimizers for LLM API calls

Package Components

OpenTrace is designed with a modular architecture:

Core Tracing (opto.trace)

  • Zero dependencies - Pure Python implementation
  • Execution graph capture - Records your workflow automatically
  • Works offline - No external API calls required

Optimizers (opto.optimizers)

  • LLM-powered optimization - Requires LiteLLM for API calls
  • Multiple LLM providers - Support for OpenAI, Anthropic, and more
  • Generative improvements - Automatically enhance your workflows

Verify Installation

Test your installation with a quick example:

from opto import trace

@trace.model
class SimpleAgent:
    def __init__(self):
        self.instruction = trace.node("Be helpful", trainable=True)

    def greet(self, name):
        return f"Hello, {name}!"

# Create and test the agent
agent = SimpleAgent()
result = agent.greet("World")
print(result)  # Output: Hello, World!

Next Steps

Start with Quick Start View Examples