What is meta-prompting?
Quick answer
Meta-prompting means using AI to write or improve a prompt, instead of writing it all yourself. You describe the job, or paste a prompt that half works, and ask the AI to turn it into clear instructions. Then you test the result like any other prompt.
Last updated
Take the free module →Module 8 of the free Better Prompts course, about 7 minutes. No account needed.
Updated · By Robert Breen
Why it matters for a small business
Models are good at spotting what a prompt leaves out: the format, the rules for missing data, the audience. Asking "rewrite this as instructions and tell me what is missing" often gets you a more complete prompt in a minute. Tools build this in too, such as the prompt generator in Anthropic's Console and Claude's skill creator.
The catch is that a polished prompt can still be wrong for your business. The AI doesn't know your prices or your rules unless you tell it, and it may add steps you never wanted. Treat the result as a draft: read every rule, test it on a lazy real request, and fix the gap the test shows.
In a real lesson: Turn a Great Prompt into Reusable Instructions
In Turn a Great Prompt into Reusable Instructions, Mike from Ridgeline Heating & Air pastes the estimate prompt he uses every time and asks ChatGPT to turn it into Project instructions: keep the rules, fill in the company, add two fixes, and list the blanks. ChatGPT writes the instructions and a list of what to send each time.
The first test shows a gap: nothing says what to do when prices are missing. Mike adds one rule, ask first. In build your own skill with the skill creator, Claude goes further: it interviews you about Brightline Painting's weekly report, drafts the skill's instructions, tests them on real crew notes and packages the skill.
Try this lesson free or read the step-by-step guide.
Common confusions
Meta-prompting vs prompt chaining
Prompt chaining splits a job into steps whose answers feed each other. Meta-prompting is a prompt whose output is another prompt.
Meta-prompting vs the research meaning
Some researchers use the term for one model coordinating other copies of itself. At work it usually means AI helping you write prompts.
Tips
- Paste your current prompt and ask what is missing before asking for a rewrite.
- Tell the AI to ask you questions instead of guessing your rules.
- Test the new prompt with a short, real request and fix the first gap you see.
Related terms
More Prompting terms
Where you use it: free lessons
- Turn a Great Prompt into Reusable Instructions (ChatGPT, 7 min)
- Build Your Own Claude Skill with the Skill Creator: A Weekly Report Template (Claude, 8 min)
Frequently asked questions
- Can ChatGPT write a good prompt for me?
- Often, yes, if you tell it the job, the audience, the format and your rules. Always test what it writes on a real example.
- Is a prompt written by AI better than mine?
- It is usually more complete, but it can't know your business. Combine its structure with your facts.
- What is a meta-prompt?
- A prompt whose job is to produce another prompt, such as "turn this into instructions for a ChatGPT Project".
- When is meta-prompting worth it?
- For prompts you reuse every week, such as a quote, a report or a reply template. A one-off question rarely needs it.