What are prompts? Why do they work? How should you evaluate them? I explore types of prompts, which ones work best in today’s LLMs.