I wonder how much of this comes from the agent picking up code quality of the project as it explores files snd works out the actions it’s taking.

Together with its inherent training becoming an average of the world. In a world where average isn’t good enough.

Or rather. Good code quality is an uphill battle you need to fight for every time you look around in the code base, to prevent the world leaking in, and the better the quality gets the more good code will the agent have in its context when it generates new code.