A prompt is the only interface between you and a large language model. Everything the model knows how to do — write code, summarize a document, reason through a problem — is accessed through the words you give it. Prompt engineering is the practice of crafting those words deliberately, so the model's output is accurate, useful, and repeatable rather than a matter of luck.
It's easy to underestimate this skill because a bad prompt often still produces something. The model rarely refuses outright; it just quietly does the wrong thing, at the wrong level of detail, in the wrong format. Prompt engineering is what closes that gap between "an answer" and "the answer you actually needed."
Over this book, you'll build a working vocabulary of prompting techniques — zero-shot and few-shot prompting, chain-of-thought reasoning, system prompts, structured output, templates, and iterative refinement — and see where each one actually earns its keep.