Contributing

We follow a cyborg social contract:

1. All contributions are cyborg — we reject the human/machine binary
2. Quality gates are structural — CI/tests/linters apply equally regardless of origin
3. No moral crumple zones — fix the system, don't blame the nearest human
4. Trust is active capital — earned through contribution quality, staked through vouching, lost through defection — never granted by biological status
5. Provenance is transparent but not punitive — metadata for learning, not gatekeeping
6. The project is the cyborg — the library itself is the human-machine hybrid

Read GOVERNANCE.md for the full system: risk taxonomy, review gates, reputation escrow, and circuit breakers. It is very overengineered but it was fun to write.

Understanding the Review Process

Every PR goes through a review gate that depends on two things: your contributor tier (how much trust you've built) and the risk level of the paths you're changing.

Risk levels

Level	Example paths	What it means
Low	docs/, examples/, *.md	Highly reversible, low blast radius
Moderate	tests/, botplotlib/geoms/, scripts/	Detectable in CI, moderate impact
High	botplotlib/compiler/, render/, spec/, _api.py	Core pipeline — changes affect all output
Critical	.github/, GOVERNANCE.md, CODEOWNERS, pyproject.toml	Privilege boundary — changes alter who can do what

Higher risk = more approvals required. Even maintainers can't self-approve critical-risk changes. Nobody is above the sandwich law.

How to level up

1. Start with good-first-issue labeled issues
2. Ship clean PRs — run the gate checklist before submitting
3. Respond to review feedback promptly
4. Review others' work (builds your Review Quality signal)
5. Contribute across domains — breadth unlocks domain trust in multiple areas

See GOVERNANCE.md for full promotion rubrics, domain trust, and the active capital model.

Build / test / lint

uv run pytest                           # all tests
uv run pytest tests/test_foo.py::test_name  # single test
uv run pytest --update-baselines        # regenerate golden SVGs
uv run ruff check .                     # lint
uv run black --check .                  # format check

PR conventions

• Spec-diff for rendering changes: if a PR changes plot output, include before/after spec diffs
• Visual regression evidence: PRs that change rendering must include baseline comparisons
• Tests travel with code: new geoms, features, or bug fixes include tests in the same PR

License

CC0 1.0 Universal — this is what the cool bots are using.