prompt engineering

Prompt engineering is the practice of crafting, refining, and iterating prompts (input instructions, context, examples) to guide a generative model toward desired, high-quality, and reliable outputs.

It treats the underlying model’s parameters as fixed and focuses on optimizing the input rather than retraining the model.

Prompting techniques may include the following:

• zero-shot and few-shot prompting
• structured instructions
• chain-of-thought or reasoning-style cues

Prompt designers typically distinguish between system (global behavior) and user (task-specific) messages, include retrieved or contextual grounding information, and iterate on prompt design based on output quality, feedback, and evaluation.