Install uv if not aready installed.

1. Clone the ARC-AGI-3-Agents repo and enter the directory.

git clone https://github.com/arcprize/ARC-AGI-3-Agents.git
cd ARC-AGI-3-Agents

2. Copy .env.example to .env

cp .env.example .env

3. Get an API key from the ARC-AGI-3 Website and set it as an environment variable in your .env file.

export ARC_API_KEY="your_api_key_here"

4. Run the random agent (generates random actions) against the ls20 game.

uv run main.py --agent=random --game=ls20

For more information, see the documentation or the tutorial video.

• available_actions to FrameData
• ACTION7 as possible GameAction

Initial Release

AgentOps is an observability platform designed for providing real-time monitoring, debugging, and analytics for your agent's behavior, helping you understand how your agents perform and make decisions.

AgentOps is already included as an optional dependency in this project. To install it:

uv sync --extra agentops

Or if you're installing manually:

pip install -U agentops

1. Visit app.agentops.ai and create an account if you haven't already
2. Once logged in, click on "New Project" to create a project for your ARC-AGI-3 agents
3. Give your project a meaningful name (e.g., "ARC-AGI-3-Agents")
4. After creating the project, you'll see your project dashboard
5. Click on the "API Keys" tab on the left side & copy the API key

1. Add your AgentOps API key to your .env file:

AGENTOPS_API_KEY=aos_your_api_key_here

2. The AgentOps integration is automatically initialized when you run an agent. The tracing decorator @trace_agent_session is already applied to agent execution methods in the codebase.

3. When you run your agent, you'll see AgentOps initialization messages and session URLs in the console:

🖇 AgentOps: Session Replay for your-agent-name: https://app.agentops.ai/sessions?trace_id=xxxxx

4. Click on the session URL to view real-time traces of your agent's execution. You can also view the traces in the AgentOps dashboard by locating the trace ID in the "Traces" tab.

If you're creating a custom agent, the tracing is automatically applied through the @trace_agent_session decorator on the main() method. No additional code changes are needed.

To submit your agent for the ARC-AGI-3 competition, please use this form: https://forms.gle/wMLZrEFGDh33DhzV9.

We welcome contributions! To contribute to ARC-AGI-3-Agents, please follow these steps:

1. Fork the repository and create a new branch for your feature or bugfix.
2. Make your changes and ensure that all tests pass, you are welcome to add more tests for your specific fixes.
3. This project uses ruff for linting and formatting. Please set up the pre-commit hooks to ensure your contributions match the project's style.

   pip install pre-commit
   pre-commit install

4. Write clear commit messages describing your changes.
5. Open a pull request with a description of your changes and the motivation behind them.

If you have questions or need help, feel free to open an issue.

To run the tests, you will need to have pytest installed. Run the tests like this:

pytest

For more information on tests, please see the tests documentation.

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