Large HA instances with many entities cause get_states to exceed the 12s
WebSocket timeout. Now loads all states via /api/states HTTP endpoint first
(much more reliable for large payloads), with WebSocket get_states as fallback.
Also bumped WS send timeout from 12s to 60s for slow HA responses.
- HA WebSocket connection with auto-reconnect
- User cards with colour coding and per-user block toggle
- Device assignment by MAC (pick from discovered devices or enter manually)
- Per-device block toggle with online/offline indicator
- Weekday/weekend schedule blocking (overnight ranges supported)
- Auto re-block on device reconnect with new IP
- First-time setup screen, credentials stored in localStorage
jessikitty