Writing and structuring model input so the output is reliable enough to use. Where the problem is what the model is given rather than how it is asked, the discipline is context engineering.
Prompt Engineering is the practice of writing and structuring a model's input so the output is reliable enough to use. It covers the instruction itself, the examples supplied with it, the output format demanded, and the order in which all of it appears.
Few-shot prompting supplies worked examples inside the prompt, specifying a task without fine-tuning. Chain of thought asks for intermediate steps before the answer and raises accuracy on multi-step work. Constraining output to a schema removes parsing failures. Placing the instruction after long reference material helps, because attention to the start and end of a long context window is stronger than to the middle. Lowering temperature narrows variation where the task has one correct form.
A prompt cannot supply knowledge the model neither holds nor retrieves; that gap closes with grounding, not with phrasing. Prompts are model-specific, and wording tuned on one model degrades on the next version. Where the problem is what the model is given rather than how it is asked, the discipline is context engineering. Any prompt carrying untrusted text is exposed to prompt injection.
