[coldtea]

One reason seems obvious/intuitive: because their own output matches their own biases, that is, is a direct result of their model walks.

[tyre]

I think this could be true if you use a single Claude context, since it has its own reasoning in the context.

But a separate code review agent does much better, in my experience.