Concept
Prompt Engineering
Practice of designing prompts that elicit reliable, high-quality model outputs.
Definition
Prompt engineering combines empirical iteration with structural patterns — role priming, few-shot examples, chain-of-thought, output schemas, delimiters — to maximise an LLM's task performance without retraining it. It is the cheapest knob to turn before considering fine-tuning.
Common use cases
- Production apps
- Evaluation harnesses
- Agent design