Hacker News Reader

Seems like the code is designed poorly if it can't be worked on in small chunks

Seems like you're dismissing the cost of artificially splitting what should be a monolith.

It outweighs the supposed productivity boost of LLMs by at least one order of magnitude if not more.