Skip to main content

Prompt engineering

Definition

Prompt engineering is the practice of crafting input text (prompts) to get desired behavior from LLMs: task format, few-shot examples, chain-of-thought, role-playing, and constraints.

It is the primary way to steer LLMs without fine-tuning: you control context, format, and examples in the prompt. Combined with RAG, prompts often include retrieved passages; with agents, they define tool use and reasoning style.

How it works

You compose a prompt (system message, task description, constraints) and optionally examples (few-shot). The LLM takes this as input and produces an output. Zero-shot uses only instructions; few-shot adds example input-output pairs so the model infers the task. Chain-of-thought (see CoT) asks the model to “think step by step” to improve reasoning. Structured output (e.g. “respond in JSON”) can be enforced via parsing or API options. Iterate on prompt wording and examples, and evaluate on a dev set to improve reliability.

Use cases

Prompt engineering matters whenever you call an LLM: it shapes behavior, format, and reasoning without changing weights.

  • Steering chat and task completion (role, format, examples)
  • Eliciting reasoning (chain-of-thought) for math or logic
  • Constraining outputs (JSON, length, tone) for APIs or UX

External documentation

See also