Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
71 lines (53 loc) · 2.13 KB

CONTRIBUTING.md

File metadata and controls

71 lines (53 loc) · 2.13 KB

Contributing to autolab

Thanks for your interest in contributing! This project provides companion tools for Karpathy's autoresearch.

Development Setup

git clone https://github.com/YOUR_USERNAME/autolab.git
cd autolab

# Install all tools in development mode
pip install -e ./auto-judge -e ./auto-steer -e ./auto-evolve

# Install dev dependencies
pip install pytest ruff mypy

# Run tests
pytest

# Run linting
ruff check .

Project Structure

autolab/
├── auto-judge/          # autojudge CLI — experiment evaluation
├── auto-steer/          # autosteer CLI — research direction suggestions
├── auto-evolve/          # autoevolve CLI — multi-agent competitions
├── skills/              # Claude Code skill definitions
├── templates/           # User-facing integration templates
└── test-data/           # Sample data for testing

Each tool is an independent Python package with its own pyproject.toml.

Making Changes

  1. Fork the repository
  2. Create a feature branch: git checkout -b my-feature
  3. Make your changes
  4. Run tests: pytest
  5. Run linting: ruff check .
  6. Commit with a clear message
  7. Open a pull request

Guidelines

  • No fake solutions — all code must work against real autoresearch data
  • Result types — use Ok/Err for error handling, never bare exceptions in business logic
  • Test behaviors — tests should validate what the tool does, not how it does it
  • Keep tools independent — each tool should work standalone with only click as a dependency
  • Backward compatible — don't break existing CLI flags or output formats

Adding a New Tool

  1. Create a directory: auto-<name>/
  2. Add pyproject.toml with hatchling build backend
  3. Add the tool's README
  4. Add a Claude Code skill in skills/autoresearch-<name>/
  5. Update the root README

Reporting Issues

Please include:

  • Which tool (autojudge, autosteer, or autoevolve)
  • Python version (python --version)
  • A sample results.tsv if relevant (anonymize if needed)
  • The full command you ran
  • The full error output