Meta-prompt generator
When you need a really good prompt — or many of them — the fastest path is often to let the model write the prompt for you. A meta-prompt does exactly that: it instructs an LLM to act as a prompt engineer, take your goal, and return an optimized, task-specific prompt. This tool assembles that meta-prompt from a few inputs so you don’t have to hand-craft the meta layer each time.
How it works
You describe the task the final prompts should accomplish, pick the output format you want those prompts to elicit, and name the target model. The generator weaves these into a structured meta-prompt that tells the model to:
- restate the task in its own words to confirm understanding,
- apply prompt-engineering best practices (clear role, explicit constraints, output format, edge-case handling),
- tailor phrasing to the target model’s conventions,
- and return the finished prompt ready to copy.
Everything is assembled locally and copied with one click. Paste the meta-prompt into ChatGPT, Claude, or Gemini and it will hand you back a polished prompt.
Tips and notes
Be concrete in the task description — “write product descriptions for handmade candles, 40 words, warm tone” yields far better prompts than “marketing copy”. For repeatable workflows, ask the meta-prompt to return a template with placeholders so you can slot in different inputs later. After you get the generated prompt, run it through the prompt optimizer or the system-prompt linter to catch any remaining weaknesses, then test it on real inputs before relying on it.