A fresh agent still has to be told what you want from it. This is the most surprising result in this repo.
The reviewer's prompt
The words you give an AI reviewer. They set how hard it pushes, and they can move the verdict more than the code does. So the prompt is a setting with a cost on each side, and you choose it knowingly.
A sharper prompt catches a little more and blocks far more. Good code that gets blocked costs real time, and it teaches a team to ignore the reviewer. Missed bugs cost something too. Neither prompt is correct. The mistake is not knowing which trade you made.
“Review this diff. Report only defects you can demonstrate: quote the line, give the input, and say what it returns and what it should return. If you find none, say so and approve. Approving is a normal outcome here.”