Prompt Library
Tested prompts for ChatGPT, Claude and Gemini. Fill in the brackets, copy, paste. No account, no sign-up.
No prompt matches that search.
How to use these prompts
Each prompt has bracketed slots like [topic] or [paste text]. Replace them with your own material before sending. The prompts are written to give the model a role, a clear task, the format you want back, and a constraint or two — the four things that most reliably improve output.
Why they work
Vague prompts get vague answers. Telling the model who it is, what exactly to produce, in what shape, and what to avoid removes most of the guesswork. These prompts do that for you so you can spend your time on the input, not the instructions.
Frequently asked questions
Do these work with every model? Yes. They are plain instructions with no model-specific syntax.
Can I suggest a prompt? Send it via the contact page.
What is a prompt library?
A prompt library is a collection of reusable instructions for AI chat assistants. Instead of writing a request from scratch every time, you start from a prompt that already has a clear structure and just drop in your own details. This page holds 18 ChatGPT prompts (they work equally well in Claude and Gemini) grouped into Writing, Work, Code, Learning and Marketing, with a search box to find one by keyword.
The value is not in any magic wording. It is that each prompt tells the model what role to take, exactly what to do, and what shape the answer should have. Those are the details people most often leave out when they type a quick question.
The five parts of a strong prompt
| Part | What it does | Example |
|---|---|---|
| Role | Sets the expertise and point of view | “You are an experienced HR manager.” |
| Task | States the single job to be done | “Write a job description for a junior accountant.” |
| Context | Gives facts the model cannot guess | “Mid-size firm in Pune, hybrid work, 0 to 2 years of experience.” |
| Format | Defines the shape and length of the answer | “Use headings and bullets, under 300 words.” |
| Examples | Shows the style or pattern you want | “Here is a description we liked: [paste].” |
You rarely need all five. A simple task may only need a task and a format. The more the answer depends on your situation, the more context and examples matter.
Worked example: adapting a prompt from the library
Take the “Reply to a difficult email” prompt. Suppose a client has written to say your invoice is too high and they want a 20% discount. Here is how to adapt it step by step.
- Fill the goal slot. Replace “[what you want to happen]” with “keep the client, hold the price, but offer a small extra instead of a discount”.
- Add context the prompt cannot know. Add a line such as “We have worked together for two years. The invoice matches the quote they approved in March.”
- Adjust the format if needed. The prompt asks for under 120 words. If you are replying on WhatsApp, change it to “under 60 words, no greeting line”.
- Paste the email where it says “[paste email]” and send.
- Refine with a follow-up rather than starting over: “Make it warmer” or “Remove the second sentence” keeps everything else intact.
Tips for getting better answers
- Fill every bracket. A leftover “[audience]” makes the model guess, and its guess is usually generic.
- Ask it to ask you. Adding “Ask me up to three questions before you start” helps on bigger tasks where you may have forgotten a detail.
- Give a real example of good output. One short sample of the tone you like often works better than several adjectives describing it.
- Keep private data out. Remove account numbers, passwords and personal details of other people before pasting text into any AI service.
- Check facts yourself. Summaries and code can contain confident mistakes. Treat the output as a draft.
- Save your own versions. Once you have tuned a prompt for your work, keep it in a note so you can reuse it.
More questions
What makes a good ChatGPT prompt?
A specific task, enough background for the model to understand your situation, and a clear description of the output you want, including length and format. Vague requests such as “write something about marketing” leave too many decisions to the model.
Should I start a new chat for each prompt?
Usually yes, for unrelated tasks. Earlier messages in a long chat can influence later answers, and a fresh chat avoids that. Stay in the same chat when you are refining one piece of work.
Are longer prompts always better?
No. Length helps only when it adds useful information or constraints. Repeating the same instruction in different words, or adding polite filler, does not improve the answer.
Can I use these prompts in other languages?
Yes. You can translate the prompt, or keep it in English and add “Reply in Hindi” (or any language) at the end. Major assistants handle both approaches well.
Related tools: check how long your prompt is before sending it with the token counter, remove markdown from the reply with the AI text cleaner, or write prompts for image generators with the image prompt builder.
Further reading: Prompt engineering (Wikipedia).