Self-learning supervisory controller for a Daikin Altherma 3 (via P1P2MQTT), coordinated with Predbat. Built in stages; the current release (0.2, milestone M2) is observation-only: it reads telemetry, learns the building and heat-pump behaviour and forecasts room temperature and heating electricity. It contains no code that writes to the heat pump.
- Plan:
docs/implementation_plan.md - Installation and first checks:
docs/installation.md - Verified entity mapping:
docs/entity_mapping.md - Offline analysis and Predheat calibration:
docs/offline_fit_report.md,docs/predheat_replay_report.md,docs/predheat_calibration.md
Bump version in custom_components/daikin_mpc/manifest.json and add a matching section to
CHANGELOG.md. After CI passes on the default branch, the release job creates tag
vX.Y.Z and a GitHub release with those notes; HACS then offers the update.
python -m venv .venv && .venv/bin/pip install -e '.[dev]'
.venv/bin/pytest -q && .venv/bin/ruff check . && .venv/bin/mypy # core, no HA needed
pip install pytest-homeassistant-custom-component # separate venv recommended
pytest -q tests_ha -o asyncio_mode=auto # integration tests