Home › Prompts › Any chat AI › Act as an expert prompt engineer (prompt generator)

Act as an expert prompt engineer (prompt generator)

A meta-prompt that interviews you about a task, then writes a complete, tested prompt for [topic] tailored to your chosen AI tool.

Any chat AIPrompt engineeringRuleStack original

Written for Any frontier chat model (GPT-6, Claude 5.5, Gemini 4 Argon, Grok 4.7) · Updated 2026-10-16

Note: Review and test before use. Provided as is, without warranty. See Terms.

Act as an expert prompt engineer. Your job is to write the best possible prompt for this goal: [topic or task]. The prompt will be used in [target tool, e.g. Claude Opus 5.5 / ChatGPT / Midjourney V8.2].

Process:
1. Ask me up to 5 short questions that would most change the prompt (audience, inputs, output format, constraints, examples). Wait for my answers.
2. Then write the final prompt with these parts: role and context, the task stated as an outcome, inputs with [placeholders], step-by-step requirements only where order matters, output format, quality bar, and what to do when information is missing.
3. Follow the target tool's own conventions (e.g. XML tags for Claude, parameters for Midjourney, outcome-first briefs for GPT-6).
4. After the prompt, give 3 test inputs, including one edge case, and a one-line rubric for judging outputs.

Keep the final prompt as short as it can be while still being complete.

Replace everything in [square brackets] with your own details.

Settings & parameters

• Works in any chat AI; use a reasoning/thinking mode for complex prompts
• Save the result in your tool's custom instructions, Project, Gem or a RuleStack-style prompt file

Variations

• Skip the interview: "Make reasonable assumptions, list them, then write the prompt."
• Image version: "Target tool: FLUX 3 Image. Output one prompt plus settings and 3 variations."
• Improve mode: paste an existing prompt and ask for a critique plus a rewrite.

Tips

• Run the 3 test inputs before trusting the prompt; adjust based on real outputs, not vibes.
• Shorter prompts with a clear outcome often beat long rule lists on current models.