Added
- Threshold alerts for device metrics. The manager already keeps a per-device history of CPU load, memory usage and temperature. It can now watch those values and send a webhook when one crosses a limit you set, and an optional second webhook when it recovers. Thresholds live under Settings → Metric alerts: CPU and memory start at 90%, temperature starts off (normal temperature differs too much between models to pick a safe default). An alert only fires after the value stays over the limit for a few readings in a row, so brief spikes don't page you. You can override any threshold on an individual device, or set it to
0there to silence that one metric. Follows the same webhook approach as the existing alerts, so it works with ntfy, Discord, or any custom endpoint.