[RistrettoMike]

I’d love to see how OpenAI’s overbuilt autocorrect would serve as a moderator in practice. Delightful chaos.

[ctoth]

I'm very very not great at prompt engineering and was able to get this working by thinking about it for 30 seconds, so I'm sure you could do way better than this[0].

[0]: https://chat.openai.com/share/69a9d595-ad26-450e-9810-da39e9...