Skip to content

Home Assistant Offline Survival Checker

Pure frontend: tick the devices and integrations you actually use, and the tool sorts them by local protocol vs cloud into three buckets - survives offline, dies offline, and remote-only while local stays fine - plus a survival checklist to do before the outage. No data leaves the browser, never reaches your instance.

The biggest operational worry about Home Assistant is whether it stays alive when the internet drops. The answer depends on whether each device or integration speaks a local protocol or sits behind a cloud. This tool buckets what you actually run by protocol: a local coordinator, on-LAN devices and cameras, local HomeKit, and local voice survive offline; vendor-cloud integrations and HA Cloud remote lose the connection. The point is to know, before the outage, which capabilities are genuinely offline-capable.

The offline survival matrix

Tick what you use and click Evaluate; the tool buckets each by protocol:

Survival checklist (do this before the outage)

  • Lock down the local path first: keep the Home Assistant host and its devices on the same LAN segment or a reachable VLAN, and do not hide local devices behind an isolated subnet that NAT blocks.
  • Prefer local protocols: if a device supports local Zigbee, local MQTT, or on-LAN ONVIF / RTSP, use that instead of the vendor cloud - a cloud path dies the moment the internet drops.
  • Make local voice an independent stack: run voice Assist with a local LLM and local TTS / STT, and do not tie it to an online cloud service, so voice still works offline.
  • Decouple remote access from local: reach the instance over Tailscale, a reverse tunnel, or HA Cloud; none of these should stop local automations when it goes down.
  • Run one outage drill: pull the WAN or stop routing, and confirm each local automation, local camera, and local voice still fires.

Honest note

This is a protocol-based static judgment: it only reasons about the categories you tick, sorted into buckets by local protocol versus cloud dependency. It does not reach your instance, does not probe the runtime, and does not connect to the internet; nothing you select leaves the browser. It helps you decide which parts to keep local before an outage; your real availability is decided by a field drill.