I'll definitely do a technical write-up in the coming days! I'm storing any valid word they type, though I'll probably quite quickly put in a TTL for words <= 3 letters long. But every thing they write is from a deterministic seeded RNG, with a tick that is randomly generated every 15 seconds, combined with the monkey salt (that I may rotate on a schedule). So you can technically go back in time and regenerate all of the content.

Supabase/postgres backend. First time for me, I actually was using firestore a week ago and migrated in a delirious frenzy at the last moment - which I think was a good call. Firebase charges per document write, and I had words as documents... which would get out of hand quickly. There's ways to be more clever than that (I also would store 16 monkeys to a single doc) but it just wasn't a good fit for what I was trying to do