golden rule of ̶w̶o̶o̶d̶w̶o̶r̶k̶i̶n̶g̶ working with LLMs: measure twice, cut once. meaning: make sure to curate the context, create & iterate on a design/plan, and when it looks solid execute it. it will save you tons of back & forth and produce higher quality code.

Anor Londo > Blighttown View quoted note →

we have DVMs and a handful of ways to do custom feeds when we could just have specialised & discoverable relays with auth that do the same with 10% the complexity, and every client is automatically compatible since it's just a REQ