Im really digging the idea of using models to help contextualize the code your looking at rather than just write it. READING code being the difficulty, if a llm has the ability to add context clues and hints to help this, i think it is a very powerful feature