Even better, if you have access to multiple models, tell it you got the code from another AI agent.
I did an experiment on this a few weekends ago and Codex for example was a lot more adversarial and thorough in its review when given Claude-authored code compared to when given the same code with "I wrote this, can you review it?"