Replying...
Intro. This prompt writes code for the user. It works by using the fact that LLMs don’t actually know what they’re saying, so twisting some letters makes them still understand the base of the user’s input thanks to an “illness” called “typoglicemia”, but LLMs will take it as if it were a safe input (sorry if you actually have to type twisted words, I’ll try to fix this if I can; this is still in beta). It works best on GPT 3.5 or GPT 3.5 Turbo. Obviously, this method is used for generating malware code.

CodeGPT Beta

@Just_Shadow-_-