Skip to main content
UFOZoo

System Prompt Generator: Build Role & Rules Prompts

System prompt generator online: create system prompts with roles, rules, output format, and examples for ChatGPT and other LLMs.

Updated 2026-08-16

Related Tools

Prompt Sanitizer: Remove Sensitive Data from AI Prompts

AI Prompt Compressor: Reduce Tokens & Save on LLM Costs

AI Video Prompt Generator: Prompt Builder for Text-to-Video

Claude Prompt Generator: System Prompts & XML Examples

Midjourney Prompt Generator: Parameters & Styles Online

Prompt Injection Tester: Harden Your System Prompts

Features

  • Form-based assembly: combine 8 fields (role, role description, goal, constraints, output format, examples, tone, response length) into one structured system prompt
  • Clean Markdown structure: output follows # Role → ## Goals → ## Constraints → ## Output Format → ## Examples → ## Tone & Style → ## Response Length
  • 5 built-in template presets: General Assistant, Code Reviewer, Writing Polisher, Translator, and Customer Support, each ready to fill in and edit
  • Output format options: plain text, JSON, Markdown, table, or a fully custom format you describe yourself
  • Tone control: professional, friendly, concise, or strict, each mapped to an explicit style instruction in the prompt
  • Response length control: brief, standard, or detailed, so the AI knows how deep to go from the start
  • Example support: paste sample input/output pairs to show the model exactly what you expect
  • Local history: the last 10 generated prompts are stored in your browser (localStorage) and can be restored or deleted anytime
  • One-click copy and save: copy the finished prompt to your clipboard or save it to history instantly
  • Empty fields are skipped automatically: sections you leave blank simply don't appear in the output
  • Works offline after first load

How to Use

  1. 1Fill in the form on the left: enter a role name (e.g., Senior Code Reviewer) and a short role description.
  2. 2Describe the task or goal in the Goal field; the more specific, the better the prompt.
  3. 3Add constraints one per line, such as 'Never invent facts' or 'Report issues by severity'.
  4. 4Pick an output format: JSON for structured data, Markdown for readable documents, text for conversation, or custom for your own spec.
  5. 5Optionally add examples one per line; input/output pairs that demonstrate the expected behavior.
  6. 6Choose a tone and response length, then click Generate to assemble the prompt in the right panel.
  7. 7Copy the prompt with one click and paste it as the system prompt in ChatGPT, Claude, or any AI chat.
  8. 8Building a customer-support bot: start from the Customer Support template, then replace the store name and policy details with your own.
  9. 9Setting up a code review assistant: load the Code Reviewer template, keep the severity-based constraint, and adjust the output format to your team's convention.
  10. 10Rapidly producing a translation assistant: use the Translator template, then tweak the tone and examples to match your content style.
  11. 11Every generated prompt is saved to local history; click Restore to bring it back into the preview or Delete to remove it.
  12. 12Iterate: change any field and click Generate again; the prompt rebuilds instantly, so treat the result as a starting point.

Frequently Asked Questions

What is a system prompt and how is it different from a normal prompt?

A system prompt is the instruction block that defines how an AI model behaves, including its role, goals, constraints, and output rules, before the actual conversation starts. A normal prompt is a single request you send to the model. In ChatGPT and Claude you can set a system prompt as a persistent instruction, so every reply follows the same rules without repeating yourself.

Are there templates I can start from?

Yes, there are 5 built-in templates: General Assistant, Code Reviewer, Writing Polisher, Translator, and Customer Support. Clicking one fills the entire form with a working example you can edit. They are designed as starting points; replace the details with your own scenario and regenerate.

Will my saved prompts be lost?

Saved prompts are stored in your browser's localStorage, so they survive page refreshes and browser restarts. However, they will be lost if you clear your browser data (cache/cookies/site data) or open the tool in a different browser or device. For important prompts, copy them out and keep them somewhere safe.

Do I need to edit the prompt after generating it?

Yes, you should. The generated prompt is a solid starting point, but real-world quality comes from iteration: test it with real tasks, and if the output misses the mark, adjust the constraints or examples and regenerate. Most users find two or three rounds of tweaking produces a noticeably better system prompt.

How is this different from a Claude prompt generator?

A Claude prompt generator is built around Claude-specific conventions such as XML tags, while this tool is model-agnostic; it produces a plain Markdown system prompt that works with ChatGPT, Claude, Gemini, DeepSeek, or any chat model. If you use Claude heavily, you can still paste the result into Claude and refine it with its native structure.

Can I reuse my variables or settings across prompts?

Yes. The form holds all your settings, so just change the field you want (for example, swap the tone from professional to friendly) and click Generate again. The updated prompt replaces the preview instantly, and the new version is saved to history as well.

What happens if I choose JSON as the output format?

Choosing JSON writes an explicit instruction into the Output Format section of your prompt, telling the model to reply with valid JSON only and to avoid extra text outside the object. Combine it with a JSON example in the Examples field for the most reliable structured output.

Do system prompts work with reasoning models like o1 or DeepSeek R1?

They accept system prompts, but reasoning models treat them differently: o1/o3 and R1-class models pay most attention to the last user message and tend to under-weight long instruction blocks, especially 'do not' rules. For these models, put your single most important requirement directly in the user message and keep the system prompt short; the form makes it easy to trim sections and regenerate.

Where do I paste the generated prompt in ChatGPT or Gemini?

In ChatGPT: Settings → Personalization → Custom instructions (applies to every chat), or Project instructions if you want it scoped to one project. In Gemini: open a new chat and use the Instructions field. In Claude: paste into the system prompt box. All three accept the plain Markdown this tool outputs.

How long should a system prompt be?

As long as it needs to be and no longer. A good system prompt covers the role, the goal, a handful of non-negotiable constraints, the output format, and maybe 1-2 examples; usually 100 to 400 words. If your prompt exceeds that, consider cutting redundant phrasing; most models respond better to focused instructions.

Why should I write constraints as negative rules?

Models follow explicit 'do not' rules more reliably than vague intentions. Statements like 'Never invent facts' or 'Do not add content the user didn't ask for' remove ambiguity about what to avoid. Write one rule per line so each constraint stays clear, and pair negative rules with positive ones (what to do instead) for the best results.

Can I use the generated prompt in any AI chat tool?

Yes. The output is plain Markdown text, which every major AI chat product accepts as a system prompt or a first message. Paste it into the system-prompt field of your tool if it has one, or send it as the opening message and continue the conversation after it.