Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
65 lines (42 loc) · 2.23 KB

CONTRIBUTING.md

File metadata and controls

65 lines (42 loc) · 2.23 KB

Contributing to Unity

Getting started

Unity depends on two sibling repositories. Clone all three as siblings:

git clone https://github.com/unifyai/unity.git
git clone https://github.com/unifyai/unify.git
git clone https://github.com/unifyai/unillm.git

cd unity
pip install uv && uv sync --all-groups

This installs Unity and its sibling dependencies (linked via [tool.uv.sources] in pyproject.toml).

Running tests

Tests use real LLM calls with cached responses. After the first run, cached responses replay instantly.

source .venv/bin/activate

# Run all tests
tests/parallel_run.sh tests/

# Run a specific module
tests/parallel_run.sh tests/contact_manager/

# Run a specific test
tests/parallel_run.sh tests/contact_manager/test_ask.py::test_name

Tests require a running backend (Orchestra). The test runner starts a local instance automatically via Docker (requires Docker Desktop). See tests/README.md for the full testing philosophy.

Code style

We use black for formatting and autoflake for unused import removal. Pre-commit hooks run automatically on commit:

.venv/bin/python -m pre_commit run --all-files

CI on forks

The full test suite requires org-level secrets (API keys, backend access). Fork PRs run lint checks only. A maintainer will trigger the full test suite on your PR after review.

Pull requests

  • Open PRs against the staging branch
  • Keep PRs focused — one concern per PR
  • Tests should pass locally before opening a PR
  • We don't require backward compatibility (see the project's aggressive refactoring philosophy)

Design principles

  • No regex routing. If the system handles something wrong, fix the prompt or tool docstring, not a hardcoded rule.
  • No defensive coding. Don't wrap things in try/except unless you're handling a specific, recoverable error. Fail loud.
  • English as API. Managers communicate through natural-language interfaces. The Actor orchestrates through English-language primitives.
  • Real LLMs in tests. We never mock the LLM client. Responses are cached for speed, not faked.

Questions?

Open a Discussion or join our Discord.