Adversarial Prompts for LLM 

Improvement via critique & does a new thread improve results?