The reason why this exists to begin with is because I was working on a simple personal project that required the use of Aphid + Firefly for API requests. Nothing unusual, until there was a whole part about "this API has rate limits" and so we returned cached results to respect the limits, again, nothing unusual. The issue only came when we were initially using xxstache, a personal project of mine out of Smiduweorc. Somehow, we ended up migrating to Redis, because we had multiple different services using the same information (xxstache uses SQLite, which we could have just used an API for reading, and technically there was no more than 1 writer, but we ended up not doing that because we started caching other stuff, which either resulted in us making multiple instances of SQLite dbs or running into locking issues). We did consider making use of Turso or something else, but we ended up using Redis since we already had a free tier instance on Aiven.
This sudden migration resulted in about 1 day's worth of rewriting, which was ok, but we all saw that this was a particularly inefficient process. As a result, after the project was done, we revisited this problem and decided that we should extract the common shapes that we didn't change and turn it into a library. Which is also why things like normalization, hashing and key derivation are outside of the scope.