Review is done by LLMs? How you guys decided to deal with prompt injection attacks?

It isn't. Doesn't involve AI. Read the post :)