@weilawei Well, there you are, then. Nothing lost by using LLMs for that task. You could probably get away with just asking it if there are bugs in the code and blindly accepting its answer too. Nobody told me I couldn't. /s @ploum