Your input please: Making emergency management information actionable in HA

Summary

Building a HACS integration for making emergency management feeds useful, and seeking your input as part of its development and progress.

Disclosure: AI used in part within repository code referred to

Consistent with now located community rules (which may in part include Want to help others? Leave your AI at the door - Blog - Home Assistant Community), the repository referred to and getting started on development in this post contains - in part - AI generated code.


After doing some work in my HA instance to bring in RSS and CAP feeds from my local emergency management agency, I started thinking a little bigger.

I want to get some input and insight from others in this area to see if what I’m thinking about is wanted, how it might be used, or what prospective users might want to see from such an integration.

How can you help?

I’d appreciate all and any input from the community on the following:

  1. Would this be useful to you?
  2. Would you be using it in the way I’ve envisioned, or some other way - and if so, what would be that be and why?
  3. Does this offer all the features and functions you’d think are relevant for a first release of such an integration?
  4. Have I missed anything of importance or relevance to this?

Where I’m at with this idea

Short version: Repo created, in part using AI generated or supported code for the baseline of what I didn’t already have for this concept and idea, and nearly ready to run initial tests on one of the initial components (maintaining and creating a directory of feeds from public sources) - after I setup a standalone HA instance for the purpose of developing, testing and validating this.

On the basis of work to successfully ingest CAP-au feeds, filter them to a location, and render any filtered results in a Markdown card; I’ve opened and started a public Github repo where I’ve developed the scaffold for getting started on this, and am almost at the stage of testing the latest iteration of the concept and approach.

If you’d like to take a look, or read more, feel welcome to by visiting:
twcau/CAP-au-for-home-assistant: Repo for future HACS package, that will enable ingest of emergency information feeds following appropriate specifications, to identify alerts relevant to one or more locations for display and action.

What does the package need to be useful?

At the moment, this is where i’ve landed:

  1. Guided setup with *my.home‑assistant deep‑links to the exact screens (Blueprint import, Developer Tools → YAML, Entities, Scripts, Events).
  2. Multi‑feed ready:
    • monitor multiple CAP/XML/geoJSON feeds (e.g., state, national, overseas agencies);
    • use per‑feed or combined views, and
    • a UI method for feed selection, supported by a list of feeds that is cached locally and updated from the repo on a regular basis (save users hunting for supported feeds for the most part).
    • validation to warn when a provided/selected feed lacks lat/long or polygons to provide sufficient useful information pertaining to where the event is or impacting.
  3. Multi‑location ready: filters the feeds for events impacting one or more specified locations and radiuses/distance from border.
  4. plug‑and‑play feed ingestion (XML), with parsing that keeps entity states short - and stores long values in attributes - (avoids the 255‑char clamp).
  5. Unified alerts sensor: a template aggregator exposes a count plus an alerts JSON attribute that dashboards and automations can consume.
  6. Dashboard card - (Markdown) listing alerts with severity, urgency, area, issued time, and links.
  7. Blueprints - for:
    • Notifications near watchpoints- (radius for circles + point‑in‑polygon for polygons).
    • Zones lifecycle- (create/update/delete dynamic zones from alert geometry).
  8. Python Scripts where needed (sandbox‑safe; no imports) for geometry purposes.

Why am I looking at this?

Whilst we already have the Global Disaster Alert and Coordination System (GDACS) integration, it focuses on the bigger events - not smaller, more impactful things that might be closer to home (say a house or factory fire close by, or a burst pipe flooding a road).

What I found, and how I think this could be helpful

So I did a bit of research on the landscape, and found that:

  • A lot of national/state/local emergency and weather agencies publish feeds for this data
  • A good number of them use common and agreed protocols for these data feeds, such as the Common Alerting Protocol (CAP) or geoJSON
  • When using these common protocols, they can include lat/long polygon information to work out the impacted
  • There was nothing I could find for HA which would allow you to:
    • Choose which feeds you monitor
    • Choose exactly where you want to monitor (lat/long or polygon, and radius or distance from same, or edge of the polygon)
    • Filter those feeds down to only those things which are within where you want to monitor

As such, it meant that a HA instance lacked information to trigger automations/actions at someone’s property so they could be alert or protect your property and family.

This automations/actions this could potentially enable might include:

  • Send a notification about an event with as much or as little information as possible;
  • Check where everyone normally at home is, and even get them to check-in by responding to a notification.
  • Automatically close doors or windows, or check if any are open and need attention;
  • Turn off air conditioners;
  • Turn on air filters/recyclers;
  • Turn on a sprinkler system for a period of time, repeatedly (say if fire or ember risk);
  • Activate a well or scheme water pump, or turn one off;
  • Top up a water tank;
  • Focus on charging up battery storage and backup;
  • Activate a foam protection system.
  • Wait for a manual human confirmation from a notification, or another trigger, before taking action to do one or more of these things.

The possibilities of things that could be done on the back of having such filtered and focused data are limited only by someone’s imagination, what equipment they can get talking to Home Assistant, and how much something might cost.

But even so much as turning on an air conditioner, activating a sprinkler system on a regular basis to deal with ember attacks, or ensuring water supply and power backup is protected are things within most users’ reach.

1 Like

Aka, the God mod? The all seeing, all knowing, all powerful add-on? All dependent on a reliable internet feed in times of disaster?
Did AI help you write all this, in a fit of hopeful grandeur?

Have you seen the bushfire.io website that my friends Downunder often refer to?

Forgive me, but I’m not sure what your reply is intending to contribute to the discussion other than playing the man and not the ball.

If your reply is intended solely to knock someone for setting out to make a decent integration and be a helper to people in the HA community, then you’ve misjudged things by a wide margin.

If your reply is intended to knock someone who knows how and makes the effort to communicate, plus has spent a good part of their working life in IT and experiencing where things go wrong, and/or use tools to help improve their writing - it’s another swing and a miss for you.

And given I’ve made clear that I’m already well on the way to achieving this - having successfully ingested CAP-au feeds for this very purpose, again - what is your point?

At this point, you’d be sent off the baseball field for three strikes.

But if your reply is to have a crack at gAI and how it can be helpful in some areas to do or achieve a thing, then find the right place for it.

Or, to riff on Mark Twain, ‘better to keep your fingers off the keyboard and appear stupid than to start typing and remove all doubt.’

As for bushfire.io, I’m well aware of it thank you, and in the context of what’s hoped to be achieved with this integration - that site, and most others like it say:

  1. If you want notifications, pay up for more than five of them.
  2. Want to use the data on a different platform (say Home Assistant), then pay up some more.
2 Likes

You want people to trust their lives and safety on a HomeAssistant add-on working reliably in times of disaster?

I’ve just spent the last week comforting my neice who watched the bushfires downunder creep close to her place, unable to turn on the pumps as the electricity was out, the diesel low, the internet was down, and the shelter she fled to only had spotty internet and cell coverage.
Do you think somebody is going to stay home and watch their HomeAssistant dashboatd, waiting for it to update, as the fires bear down on them, burning them to a crisp?

I’d like Home Assistant to be an additional support for - or an extra tool in the toolbox if you will - for people in times of approaching emergency.

Something which is made clear, plain as day, in the original post.

The people this integration targeted at will already most likely have a good comprehension of the risks, and the need to ensure primary source material is what drives their response.

And good thing I’ve already countenanced the possibility of people who may not be as well informed in the design and planning of the integration - as I’m also including:

  • a strong disclaimer and clear warning language,
  • a clear click wrap disclaimer during installation, so that anyone using the integration is fully informed and aware before they go ahead and install, along with a record to document they’ve accepted same, and
  • safety checks and repair warnings that stop the integration from working if someone tries to alter the code or function to suggest they didn’t accept the click wrap disclaimer, or the integrations’ disclaimer gets updated/revised
  • prompt signals to Home Assistant’s Repairs function if a feed, or the integration stops working, including the start of ensuring these signals are easily translated into multiple languages.
2 Likes

Continuing the discussion from Your input please: Making emergency management information actionable in HA:

I’m relatively new to HA, just this year. I’m finishing non-functionals like battery and system notifications.

As I’m finishing, I have an eye towards what could I do in this new world which is a lot.

I think this is a good idea. I have a yolink device that can speak warnings I record

  • I live in an area the can have tornados and sever weather
  • If there is police or safety activity in the area

So I could define emergency info and create related yolink spokens

As for AI, I don’t care. If it’s written and tested for function, purpose that’s what matters to me. No matter who, or what, writes it… it’s on me to test its integration.

I wouldn’t use it right away, but I would look at it and considering its integration

Do we go further, the Minnesota ICE alert (and Minnesota ice alert, both important for some right now) addon, the heart pacemaker addon?
What if it breaks and somebody dies? Who ya going to sue? Nabu Casa? The forum mods? The GitHub poster? Claude? ChatGPT? Point to your extensive disclaimers? The lawyers are already rubbing their hands in glee, class action and wigs at ten paces at the ready.
Yeah/naah.

I’m not attacking any poster, and aware they have - I just think it is not a good idea, all your eggs in one basket.

I’ll stick to my battery powered torch and radio, and listen to the designated emergency warning channel, taking the advice of trained emergency personnel, thank you.

I’m not trusting that some add-on, created by an unknown person, using tools that are trained on unreliable data, getting data over an unreliable connection, to save my life. I’m invoking my tested and up to date emergency plan, grabbing my valuables, family, pets, and getting the heck out of there to safety if they tell me to evacuate, and down in the shelter if they tell me to stay.

You make some very wild assumptions, that this integration would somehow replace not only any other emergency communication methods existing, but that it would remove its users brains somehow.

Personally I don’t see any kind of a legal risk here, or any other kind of risk. Tools like this area very obviously complementary.

Seriously, it’s like you would complain TV stations transmit information about emergencies, because you can’t watch TV when there is no power.

3 Likes

Awesome. Thanks. Ignore the strong negative bias from some :smiley:

1 Like