A modular, micro-service-oriented toolkit that ingests raw Home Assistant telemetry, detects anomalies, reasons about context with LLMs, and proposes (or executes) remediations.
| # | Folder | Purpose | Status |
|---|---|---|---|
| 01 | components/01_ingestion |
Integrate Home Assistant event stream, MQTT topics, and REST pulls into a unified message bus. | ✅ Done |
| 02 | components/02_time_series_store |
Optimised time-series DB & feature store. | ✅ Done |
| 03 | components/03_feature_engineering |
Data cleansing, resampling, and feature derivation. | Planned |
| 04 | components/04_anomaly_detection |
Rule-based & ML models run hourly to flag anomalies. | Planned |
| 05 | components/05_media_analysis |
Optional worker for image/video understanding. | Planned |
| 06 | components/06_llm_reasoner |
LLM-powered explanation & action planning. | Planned |
| 07 | components/07_notification_ux |
Push notifications and feedback capture. | Planned |
| 08 | components/08_automation_executor |
Executes approved actions via Home Assistant services. | Planned |
| 09 | components/09_model_training |
Offline model training & evaluation pipelines. | Planned |
| 10 | components/10_observability |
Metrics, logs, and health dashboards. | Planned |
| 11 | components/11_devops |
CI/CD pipelines and deployment manifests. | Planned |
| 12 | components/12_docs_config |
Centralised docs, schemas, and config templates. | Planned |
📚 High-level architecture lives in architecture/project_overview.md.
🔖 Current sprint focus lives in .llm_context/global_status.yml.