Install python -m pip install -e ".[dev,dashboard]", then run python -m pytest,
python -m ruff check . and python -m ruff format --check . before opening a PR.
Use SI units in model code and include equations or a primary reference for physical changes.
Keep controller parameters fixed across scenarios. Do not change published numbers by hand:
run python scripts/reproduce.py after changes that affect simulation behavior.
A new controller needs a fresh-state factory entry, the common Measurement interface,
explicit actuator handling, a nominal closed-loop test and a documented mathematical law.
A new plant model needs an independent analytical limit or convergence test.
Report software bugs with the exact command, seed, run.json and the observed result.
Discuss model assumptions and controller tuning in issues; negative benchmark results are useful.