Nick Perry tests five prompt modifiers claimed to change Claude's behavior: /human for human-sounding responses, el10 for simplification, x10think for extended reasoning, killcritic to bypass agreeableness, and alt3 for multiple answers. None of them actually work.
Steal these 5 secret codes for Claude. Number 1, /human. Put this before your prompt and you'll get something that sounds a lot more human. Number 2, el10. It makes complex topics very easy to understand. Number 3, x10think. It forces Claude to think longer before it answers. Number 4, killcritic. It stops it from agreeing with everything you say and gives the ugly truth. Number 5, alt3. Gives you 3 completely different answers to pick from. So never ever use ChatGPT because
