I’ve been thinking about a format where you are given a problem to complete within a time constraint. You turn in git history and prompt history. Seems to duplicate what actual coding is starting to look like.

Really nice! Differentiates between the ones you should pay attention to vs ones that are not executed. How reliable would you say that differentiation is?