Comment by petra

Given the limits of LLM's regarding context, why is zero dependencies the goal ?

Shouldn't something like maximum reuse/abstraction, minimal written code, should be a better goal ?

Possible leading to more llm-scalable systems ?