I agree, I spent far too long figuring out where the actual prompts are and assumed they were just gatekeeping like you see in the art generation scenes until I

I agree, I spent far too long figuring out where the actual prompts are and assumed they were just gatekeeping like you see in the art generation scenes until I saw the nb link. Safe/sane LLM usage 101 means knowing what you are sending to the model, I personally would never summon some unknown 3rd party prompt. For example, the gangster one has a specific instruction not to condone violence, while this is good for safety, it would be confusing if you didn't know the actual prompt when the gangster starts lecturing you and one may assume it is an OpenAI phenomenon and not built into the summoned prompt.