Thanks, I really appreciate that. Privacy is the hill I am building this on. If you have a quick instinct on the crisis button, I would love it. Should it stay fully local as a grounding checklist with quick contacts, or is there something you would expect it to do that still fits a privacy first tool.
I'd think local contacts make sense. My instinct is also to say people should be able to add personal contacts they'd want to connect with in times of crisis (because when you're IN crisis you can't think of this, but when you aren't maybe you can) though of course I can see how that could make SOME people feel worse depending on wording.
If anyone wants the implementation details: the default mode is local-first storage with offline support (PWA). I’m deliberately trying to avoid “health data as telemetry.” Network usage is feature-scoped (e.g., correlation features only when enabled). If there are specific threat-model questions you want answered (“what do you store”, “where”, “how do exports work”), ask and I’ll be precise.