About this tool
A good AI prompt is most of the work: telling ChatGPT, Claude or Gemini to "act as" a specific role — a Linux terminal, a senior code reviewer, a SQL database, a blunt editor — before asking your real question consistently produces sharper, more useful answers than a bare question on its own. This library holds 525+ of those ready-made role prompts. Search by role name or by a keyword from the prompt itself, filter by rough category, then copy a card straight into your AI chat of choice as your first message.
Most prompts include placeholders like ${topic} or [my placeholder] — replace those with your own details before sending, or just send the prompt as-is and answer the model's follow-up questions. Nothing here is tied to a specific AI product: it is plain natural-language text, so it works the same way in ChatGPT, Claude, Gemini, Copilot or any other chat interface. Everything runs and is searched locally in your browser.
This library is built on top of the open-source, public-domain (CC0) prompts.chat community project. That upstream dataset has grown to thousands of contributions of very uneven length and quality, so the set here is a curated subset — chosen for being genuine, reasonably concise, English-language "act as X" prompts rather than pasted documents or image-generation prompts.
Frequently asked questions
What exactly is this prompt library?
It is a searchable collection of pre-written "act as X" prompts — short briefs that tell an AI chat tool to take on a specific role (a Linux terminal, a copy editor, a SQL database, a career coach, and hundreds more) before you start typing your actual request. Instead of writing a detailed instruction from scratch every time, you find a matching card, copy the text, paste it as your first message, and the AI responds in that role for the rest of the conversation. It saves the trial-and-error of prompt writing for common, well-tested use cases.
Can I use these prompts with any AI chatbot — ChatGPT, Claude, Gemini?
Yes. These are plain-text instructions in natural language, not code tied to one product, so they work the same way in ChatGPT, Claude, Gemini, Copilot, or any other AI chat interface: paste the prompt as your message and the model follows it. That said, models differ in how strictly they hold a persona over a long conversation and in exact behavior for edge cases like role-play or "jailbreak"-style prompts — treat every prompt here as a strong starting point to adapt, not a guarantee of identical output across every AI.
Are these prompts free to use, even commercially?
Yes. The underlying dataset is released under CC0 — public domain — meaning the original authors have waived copyright and you can copy, adapt, remix and use these prompts for any purpose, personal or commercial, without asking permission or crediting anyone. We still credit the source project below as good practice, but there is no legal requirement to do so on your end.
Where does this data come from, and why do some categories have more prompts than others?
This library is built on top of the open-source, community-maintained <a href="https://prompts.chat" target="_blank" rel="noopener noreferrer">prompts.chat</a> project (also known as <code>f/awesome-chatgpt-prompts</code>), where anyone can submit a prompt via a public pull request. It has grown into a large, unevenly-sized dataset — thousands of submissions ranging from short classic personas to huge pasted documents and image-generation prompts — so we curated a subset of genuinely reusable, reasonably concise text prompts and sorted them into rough categories using simple keyword matching, not manual review of every entry. That is why the mix skews toward whatever the community happened to contribute most (developer and coding personas are especially common), and why a handful of prompts land in "General" when they don't clearly match a category.