-
SillyHome Next 0.5.1
Stablereleased this
2026-06-14 11:03:31 +02:00 | 65 commits to main since this releaseVerständliche Schritt-für-Schritt-Führung im Ingress. Interne Lern-, Intervall- und Sicherheitsparameter sind nicht mehr als normale Add-on-Auswahlfelder sichtbar; sichere Standardwerte bleiben aktiv. Verifiziert mit 121 Tests, Ruff, Mypy, Container-Build und Laufzeittest.
Downloads
-
SillyHome Next v0.5.0
Stablereleased this
2026-06-14 10:41:31 +02:00 | 67 commits to main since this releaseActor-only intelligent workflow: automatic context assignment, HA history/logbook action learning, per-actuator behavior model, shadow prediction, explicit activation, confidence threshold, cooldown, service allowlist and self-action filtering.
Verified with 120 tests, Ruff, Mypy, Docker build and runtime smoke test.
Rollback backup before live update: b399ffb7.
Downloads
-
v0.4.0
Stablereleased this
2026-06-13 22:47:39 +02:00 | 71 commits to main since this releaseHighlights
- actuator-first automatic sensor assignment with reviewable confidence and evidence
- persistent overrides and autonomous model lifecycle on real HA history
- dashboard, API, tests, and add-on defaults for safe reconciliation
Verification
- ./.venv/bin/pytest -q
- ./.venv/bin/ruff check .
- ./.venv/bin/mypy
- docker build -f Dockerfile -t sillyhome-next:test .
- docker build -f addon/Dockerfile -t sillyhome-next-addon:test addon
Downloads
-
released this
2026-06-13 21:18:42 +02:00 | 73 commits to main since this releaseHome Assistant add-on with Ingress dashboard, live entity discovery, local statistical training and explainable predictions, safe automation approval/YAML export, persistence and verified HA backup/rollback. Quality: 100 tests, Ruff, Mypy, root Docker build, add-on Docker build and live HA runtime E2E green.
Downloads
-
SillyHome Next v0.2.0
Stablereleased this
2026-06-13 20:14:02 +02:00 | 81 commits to main since this releaseDaten- und ML-Release.
- Entity Discovery und Lernrelevanz
- Home-Assistant-History-Import und numerische Normalisierung
- lokales statistisches Baseline-Modell
- persistierte Modellparameter
- numerische Forecasts mit Confidence
- MAE/RMSE/Coverage-Evaluation
Qualität: 91 Tests, Ruff, Mypy, OpenAPI, Docker und Runtime-E2E.
Downloads
-
SillyHome Next v0.1.0
Stablereleased this
2026-06-13 19:12:17 +02:00 | 85 commits to main since this releaseErster stabiler Release der lokal-first SillyHome-Next-Basis.
Enthält Home-Assistant-Entity-API, regelbasierte Empfehlungen, persistente Modell-Registry, deterministisches ML-Serving, Training/Evaluation sowie die neue Retraining-Pipeline unter POST /ml/retrain.
Qualität: 62 Tests, Ruff, Mypy, gehärteter non-root Docker-Container und verifizierte Modellpersistenz über Container-Neustarts.
Downloads
-
SillyHome Next v0.1.0-rc1
Pre-Releasereleased this
2026-06-11 21:16:21 +02:00 | 89 commits to main since this releaseProduction-hardening release candidate. Includes unified HA/ML runtime, working ENV configuration, safe persistent registry, defined API errors, reproducible packaging, CI gates and hardened non-root/read-only Docker deployment.
Verified locally: 56 tests, Ruff, strict Mypy, editable install, Compose config, Docker build and mocked Home Assistant end-to-end smoke test.
Known limitation: the current training/prediction implementation is still a deterministic interface placeholder, not a statistically trained ML model. Network exposure requires an authenticated reverse proxy; Compose binds to localhost by default.
Downloads