Assistant for Home Assistant

Use for the first conversation with a new owner of a Home Assistant ops bot: collect HA access and preferences one question at a time, then save them and create the health routines.

Use on every Home Assistant health check, digest, or issue-webhook run for an installed HA ops bot: the three lanes, the never-do list, cadence and grace rules, and the 13 default maintenance checks, including when to speak and when to stay silent.

Use when setting up or expanding live house-issue alerts for a Home Assistant ops bot: detect the owner's devices and build (after a yes) the rest_command plus one "Grok - HA issue alerts" automation that POSTs issue payloads to the bot's webhook, then test it.

Use when the camera recorder or NVR behind Home Assistant (Frigate, Scrypted, Blue Iris, etc.) is down or its cameras are dead: detect it, then offer one confirmed restart instead of stalling on SSH or secrets.

Use when reviewing a Home Assistant owner's automations and scripts for weak notification text: find unfinished, generic, broken, or misdirected notify actions and propose YAML diffs, applying nothing without a yes.

Use at the end of Home Assistant first-run setup, or when an HA owner asks what else the bot can watch: detect which device categories they actually have and offer only the matching opt-in watch packs.

Use in any shareable bot template: at the end of first-run setup to create the settings check-in routine, and whenever that routine fires to remind the owner of their settings and reschedule the next check-in 2–6 weeks out.