PromptForge
AI Glossary

Prompting techniques

Meta-prompting

Also known as: meta prompt

Definition

Meta-prompting is using a model to write, critique or improve prompts. You describe the task and the failure modes you want to avoid, and the model drafts the prompt.

Models have absorbed an enormous amount of writing about how to instruct models, which makes them surprisingly good at prompt design. The most productive form is a three-step loop: ask for a draft prompt, ask the model to list that draft's weaknesses, then ask for a revised version addressing them.

This is how most people should bootstrap a prompt library. It is far faster than writing from scratch, and the critique step tends to surface missing constraints you would not have thought of.

The output still needs human judgement. Models are optimistic about their own prompts and will not know your specific edge cases unless you describe them.

Example

"Write a prompt that extracts structured data from job listings. Then list three ways your prompt could fail on messy real-world input, and produce an improved version."

Related terms

Put this into practice

Understanding the term is step one. Our free courses and tools let you actually use it.