reverse engineering code is still pretty average, I'm fare limited in attention and time but LLM are not pulling their weight in this area today, be it compounding errors or in context failures.