This is routine. We have Gemini (which is not our coding model) review our PRs and it genuinely catches mistakes. Even using the same model as the creator, without its context to bias it, would probably catch many mistakes.