Future coding agents could lean into this, and instead present the entire context to the user in a single screen. Rather than simply append to it, you could go back, add, remove, modify to a far more complex representation. A graph? A tree? Not sure.