Negative Prompting

Your LLM keeps adding disclaimers. Asking for 'clean output' doesn't help. What technique explicitly states unwanted behaviors? Flag format: CONGRESS{[technique_description]} Example: CONGRESS{repeat_instructions_twice}