A universal prompt that turns any topic into a MAXIMAL, complete answer. Works with any LLM chat interface. Starts with a numbered question round, supports SK!P commands to skip questions, and delivers deep, structured output on any topic.
Prompt | License | Contributing | Changelog | Security
Works with any chat interface that lets you send a text message:
- ChatGPT
- Claude
- Gemini
- Grok
- Mistral
- DeepSeek
- Perplexity
- Local models (Ollama, LM Studio, etc.)
- Any other chat-based LLM
The only requirement: you can paste text into the chat window.
Important: This prompt is too long for most "Custom Instructions" fields. Instead, paste it directly into the chat window as your first message.
- Open your preferred LLM chat.
- Start a new chat.
- Open pr0mpt.md in this repository.
- Click the copy icon in the top-right corner of the code block.
- Paste the entire prompt into the chat window.
- Send it.
- Now write any topic in the next message.
That's it. The prompt will now run for the rest of the conversation.
- Custom Instructions fields - too long
- System prompt fields - not needed
- Memory or similar fields - wrong place
It belongs in the chat window, as your first message.
The prompt turns any topic into a MAXIMAL, complete answer in three phases:
- Question round - The model asks 6-12 numbered clarifying questions before producing any content.
- Skip commands - You can skip one, several, or all questions.
- MAXIMAL delivery - Once the question round is done, the model delivers a complete, structured answer with no length limit.
| Command | Effect |
|---|---|
SK!P |
Skips the next question |
SK!P 2 |
Skips the next 2 questions |
SK!P 3 |
Skips the next 3 questions |
SK!P all |
Skips all remaining questions |
SK!P [No] |
Skips question number X specifically |
Skipped questions get a sensible assumption, which the model marks
with [ASSUMPTION] in the evaluation step.
| Level | Content | Extras | Length |
|---|---|---|---|
| Little | Minimal answer | No | Short |
| Medium | Solid answer | Few | Medium |
| Lots | Detailed | Yes | Long |
| MAXIMAL | Everything, full | All | Unlimited |
- Works with any LLM chat interface
- Numbered question round before every answer
- Skip commands: SK!P, SK!P 2, SK!P 3, SK!P all, SK!P [No]
- Explicit [ASSUMPTION] tracking for every skipped question
- MAXIMAL mode: no length limit, full structure
- Language chosen by the user at the start of the conversation
- No emojis, plain Markdown only
Which LLMs does this work with? Any chat-based LLM where you can paste text: ChatGPT, Claude, Gemini, Grok, Mistral, DeepSeek, Perplexity, local models, and more.
Why can't I put this in Custom Instructions? The prompt exceeds the character limit of most Custom Instructions fields. Paste it into the chat window instead.
Do I have to paste it every time? Yes. Every new chat starts fresh, so paste the prompt at the start of each new conversation.
Does it work on mobile? Yes. Copy the prompt from pr0mpt.md, open your LLM app, paste it into a new chat.
What if I do not want any questions?
Write SK!P all -> MAXIMAL after your topic.
Why no emojis? Because the output is meant to be clean, copy-pasteable, and readable in any context.
See CONTRIBUTING.md for how to contribute.
This project is source-available, not open source.
You may use it for personal, educational, and internal business purposes. You may modify and adapt it for your own use.
You may NOT redistribute, publish, share, sublicense, or resell this prompt in original or modified form.
See LICENSE for the full text.
See CHANGELOG.md for version history.
See SECURITY.md for how to report issues.