zlacker

It’s the inconsistency that gets me. Very similar tasks, similar complexity, same code base, same prompting:

Session A knocks it out of the park. Chef’s kiss.

Session B just does some random vandalism.