when you really eat the meat of the results of a well designed LLM coding agent and have done all the things that are in your low hanging fruit list, it makes you pause.

yes, this tech is amazing, learn to use it now you won't regret it.

then you need bigger problems. like ones you kinda solved almost before, but ran into funding and lack of vision problems.

i have found one i really wanted to sort out. an archive for nostr, that stores everything except ephemeral events, and then, make this a primary data source for standard running instances, such that they fetch and cache results from archival relays and maintain storage limitations, based on what is actually made available to the "cache" relay.

currently implementing this now, since i have an archive i'm running, which is going to spend forever eating the whole nostrwebs for all events it can find on all relays.

soon, my favourite people, who i follow, will be able to query my relay and find anything from the archive, and share it, and other users will be able to find the old diamonds as well. public readable is a thing. user driven is something else that i think is far more important especially when it makes that search resulut available. i'm gonna think about how to leverage this for motivating people to zap to pay for it. and how that looks, also. i am still always and ever bearish on ecash but i think i'm gonna comfort this system because i think that makes people more able to understand how i'm using it in better ways (such as for a generic access token without identifdication for a relay).

hmm would be kinda cool to have it automagically set up a hidden service too. why not.