unifi-protect

You can retrieve UniFi Protect camera snapshots and smart detections by querying recent events from the websocket buffer or specifying a historical time range. The skill fetches inline image snapshots and detailed detection data without manual UI navigation.

You need a UniFi Protect MCP server with your host address and admin credentials to connect and manage cameras. This setup allows the skill to retrieve events, list cameras, and fetch RTSP stream URLs for playback.

Yes, you can control UniFi Protect hardware settings including PTZ zoom presets, lights, chimes, and alarm arming. Mutations require explicit opt-in environment variables to enable write permissions and use a preview-then-confirm flow for safety.

The safety model gates all mutations behind explicit opt-in environment variables that enable write permissions. Changes to alarm states and Known Faces updates, merges, or deletes follow a preview-then-confirm flow to prevent unintended hardware modifications.

The best way to review detections is to ask the AI to list the newest smart detections and include a snapshot for the relevant camera. This retrieves real-time event data, smart detection details, and inline image evidence quickly.

You cannot make changes to recording settings or chimes because all hardware mutations are disabled by default. You must explicitly set opt-in environment variables to enable write permissions before the preview-then-confirm flow allows any modifications.