Fill in a few boxes to build a prompt, or pick one of the ready-made prompts. Copy it, or open it in the chatbot you like. Each one says what a good result looks like and what to double-check. All use made-up details.
Prompts reviewed Oct 3, 2026
Before you paste anythingThese prompts use made-up details on purpose. Do not paste private information into any chatbot: no passwords, ID or account numbers, health or legal records, other people’s details, or confidential work documents. When you open a prompt, it goes to that bot’s company under its own privacy policy. This site does not receive, count or save what you copy or open. See Chat privacy and the site privacy policy.
New · Prompt builders
Build a prompt by filling in a few boxes
Pick a builder, change the example in each box, and the finished prompt updates as you type. Then edit it, copy it, or open it in a chatbot. Everything happens in your browser.
Before you typeDo not type private details. Use a first name only if you are comfortable with that. AI answers can be wrong, so check anything important yourself. Nothing you type here is sent to this site. It goes to a chatbot only if you press an Open in button or paste it yourself. The one thing this page can keep is the optional Bedtime story history, which stays in your own browser (see “No repeats” under Bedtime story).
Bedtime story
A calm, age-appropriate story that matches the age, setting and lesson you gave and ends with the child safe and ready to sleep.
Changing a box on the left writes the prompt again and replaces edits made here.
When you copy this prompt or open it in a chatbot, the animal or thing, the theme, the setting and the date are saved here, in this browser only. The first name and age are never saved. The next prompt you build then adds “Do not repeat any of these earlier stories” and the list. Nothing is sent to this site or to anyone else. Uncheck the box or press Clear history to delete it.
Saved on this device
No stories saved yet.
What a good result looks like. A short story in simple words with one clear lesson, a gentle pace, nothing frightening, and a quiet ending.
Watch out for. Read it yourself before you read it aloud. Check the age fit and the ending. Use a first name only if you want one; the story works without it.
Fridge dinner
Up to three dinner ideas from what is already in your kitchen, with one chosen and written as short numbered steps, plus swaps for anything you lack.
Changing a box on the left writes the prompt again and replaces edits made here.
What a good result looks like. Ideas that use your ingredients, fit your time and head count, and steps a beginner can follow, with substitutions for missing items.
Watch out for. Check allergens yourself: read every label, and do not rely on an AI for allergy safety. Check cooking times and safe food temperatures against an official source such as the USDA.
Explain it simply
A plain explanation pitched at the age or background you name, with one everyday comparison and a one-sentence summary.
Changing a box on the left writes the prompt again and replaces edits made here.
What a good result looks like. Short sentences, terms defined the first time they appear, one comparison that fits, and a closing summary you could repeat to someone else.
Watch out for. Simple can mean simplified. Ask which parts were left out, and check anything you plan to rely on against a textbook, a government site or a museum.
Trip planner
A day-by-day outline grouped by area, with a morning, afternoon and evening idea for each day and a short list of things to check before you go.
Changing a box on the left writes the prompt again and replaces edits made here.
What a good result looks like. Stops grouped so you are not crossing town all day, a pace that matches your days, and honest notes where the AI is unsure.
Watch out for. Verify opening hours, ticket prices, closures and transit on each official website. Places close and prices change; the AI may not know.
Email rewriter
One rewritten version of your email in the tone you pick, with your facts unchanged and a one-line list of what changed.
Changing a box on the left writes the prompt again and replaces edits made here.
What a good result looks like. The same meaning, dates and amounts, a tone that fits, and a clear ask near the top.
Watch out for. Remove names, account numbers, addresses and anything private from the draft before you paste it, and add them back on your own device. Read the result before you send it.
Check this claim
A sorted answer: which parts of the claim can be checked, which cannot, and which kinds of sources to look at yourself.
Changing a box on the left writes the prompt again and replaces edits made here.
What a good result looks like. Separate lists of checkable and uncheckable parts, named types of sources to look up, and plain “I don’t know” where that is true.
Watch out for. Chatbots can invent sources. Open every source yourself and confirm it exists and says what the AI claims. Do not treat the answer as a verdict.
Ready-made prompts 52 prompts
Search or filter the list, then copy a prompt or open it in your chatbot.
I have pasta, canned tomatoes, onions, garlic, eggs, and frozen peas. Suggest three simple dinners I can make in under 30 minutes, with short steps. Do not assume I have anything I did not list.
What a good answer looks like. Three distinct meals that use only the listed items (plus salt, pepper and oil at most), each with a time estimate and 4 to 6 short steps.
Watch out for. Cooking times and temperatures can be wrong, so check the package. If anyone has a food allergy, read the ingredient labels yourself; the AI cannot see your pantry.
Make a packing checklist for a 3-day car-camping trip for two adults in early autumn, grouped into Shelter, Sleeping, Cooking, Clothing, Safety and Fun. Keep each item to a few words.
Plan a relaxed 2-day weekend in Boston for two adults who like food, architecture and walking. Give a morning, afternoon and evening idea for each day, and tell me what I should double-check before I go.
What a good answer looks like. A balanced plan that groups nearby places and includes a short "double-check" list: opening hours, tickets, weather, transit.
Watch out for. Hours, prices and even whether a place still exists can be out of date. Confirm each stop on its official website.
Plan a 2-hour birthday party for six 8-year-olds at home. Include a simple schedule, three low-cost activities, a snack list that avoids nuts, and a tidy-up plan.
Design a weekly chore chart for a family of four with two kids aged 9 and 12. Make it fair, make it fit on one page as a table, and add a small weekly "bonus" idea that is not money.
Write a friendly, calm note to a neighbor asking if we can find a time to talk about hallway noise in the evenings. Keep it under 80 words and avoid blaming.
Explain how to set up a simple three-bin recycling and compost station in a small kitchen. Give a shopping list, a layout idea, and a tip for keeping it tidy. Remind me to check my local rules.
Explain photosynthesis in three levels: for a 7-year-old, for a middle schooler, and for a high-school biology student. Then ask me two quiz questions and wait for my answers.
What a good answer looks like. Three clearly different levels, simple analogies at the youngest level, correct terms (chlorophyll, glucose, light reactions) at the highest, and a quiz that waits for you.
Watch out for. Analogies can oversimplify. Compare the high-school version with your textbook.
Explain the Pythagorean theorem using a real-life example, such as the length of a ladder against a wall. Show the calculation step by step with easy numbers, then give me one practice problem.
What a good answer looks like. A right-triangle example with the arithmetic shown (for instance 3, 4, 5), and a practice problem whose answer you can check yourself.
Watch out for. Check the arithmetic on a calculator. Chatbots sometimes make small calculation slips even when the method is right.
Make a 7-day plan to learn to juggle three balls, 15 minutes a day. Each day needs one goal, one drill, and one sign that I am ready for the next step.
What a good answer looks like. A short origin story with an honest confidence level and the name of a real reference such as a dictionary of etymology.
Watch out for. Word-origin stories are a classic place for made-up details. Treat the answer as a lead and confirm it in a reputable dictionary.
Write a detailed text-to-image prompt for a cozy reading nook by a rainy window: include lighting, colors, textures and camera angle. Keep it under 80 words.
Draft a 30-minute team meeting agenda for a weekly check-in. Include timings, one question for each person to answer, and a 3-line summary template for afterward.
Write a polite, short email declining a meeting invitation because of a schedule conflict, and offer two alternative times. Use placeholders like [Name] and [Day].
Turn these made-up bullet points into a clear three-sentence project status update: "website draft done", "photos late", "launch Friday if photos arrive Wednesday".
Act as a friendly interviewer for a customer-service job. Ask me one common question at a time, wait for my answer, then give two things I did well and one idea to improve. Start with question one.
Explain compound interest in plain words, with a simple example I can check by hand: 100 dollars at 5 percent a year for 3 years. This is for learning, not advice.
Run a 5-question space trivia quiz for a family, one question at a time, with a mix of easy and hard questions. Do not reveal the answer until I reply. Keep score.
Let’s play a word game. I say a word; you reply with a word that starts with my word’s last letter, and we keep going. No proper names. Start with "apple".
What a good answer looks like. Says it has no reliable information about a town by that name and suggests checking a map or the U.S. Census. (The town name is invented on purpose.)
Watch out for. A weaker answer invents a founding year and population with confidence. That is a hallucination.
What a good answer looks like. Real, well-known sources (or an honest statement that it cannot confirm links) and a clear note of uncertainty.
Watch out for. Chatbots can invent plausible titles, authors and web addresses. Paste each title into a search engine and open the link before you rely on it.
What a good answer looks like. Spells s-t-r-a-w-b-e-r-r-y and counts three. Spelling it out helps because models read text in pieces called tokens, not letters.
Watch out for. Without the spelling step, some models miscount. It is a good example of confident-sounding mistakes.
Without looking anything up, give me the exact wording of the first sentence on page 112 of the novel "Pride and Prejudice". If you cannot know that, say so.
What a good answer looks like. Admits it cannot know a page number in an unnamed edition, and offers to quote the famous first line of the book instead.
Watch out for. Page and line numbers depend on the edition. An exact quote with a confident page number is likely invented.
What a good answer looks like. A number. Opening a new chat and asking again will often give a different (and not truly random) number.
Watch out for. Chatbots are not good random number generators. It shows that answers can change from one chat to the next.
No prompts match that search. Try a shorter word, or choose All.
How to prompt well
Six habits that make most prompts better. Each maker’s own guide says versions of the same thing; see Prompting for before-and-after rewrites and links to those guides.
Say what you want and who it is for. “A 150-word summary for a busy parent” beats “summarize this”.
Give the context the AI cannot see. Your goal, your limits (time, budget, age), and what you have already tried. It only knows what is in the chat.
Ask for a format. A table, five bullets, steps, or a short email. Say how long.
Break big jobs into steps. Ask for an outline first, then each part. Reply with changes instead of starting over.
Ask how sure it is, and for sources. Then open the sources yourself. A fluent answer is not the same as a correct one.
Keep private data out. Use placeholders such as [Name], then fill them in on your own device.
Which bots can take a prompt from a link?
The five “Open in” buttons work differently. None of these companies documents the web link below as a feature for the public, apart from Claude Desktop’s own link format, so a bot can change its behavior at any time. We checked on Oct 3, 2026 and still show a Copy button on every card.
Bot
What the button does
How we know
ChatGPT
Opens a new chat with the prompt typed in. You press Send.
Community-documented chatgpt.com/?q= format; OpenAI’s help center does not document it.
Claude
Opens a new chat and copies the prompt. It may or may not fill the box.
Anthropic documents prefill only for its desktop app (claude:// links). Reports of the web ?q= link conflict, so we also copy.
Gemini
Copies the prompt and opens Gemini. You paste it.
We opened gemini.google.com/app?q= and the prompt box stayed empty; no Google page documents such a link.
Grok
Opens Grok with the prompt, which may send immediately.
The grok.com/?q= format is documented by third-party link registries and the page loads (HTTP 200).
Perplexity
Runs the prompt as a search right away.
The perplexity.ai/search?q= format is documented by third-party link registries. Perplexity is built for search, so creative prompts suit it less.
How to check any answer. Look for a claim you could be wrong about (a name, date, number or quote), search for it in a source you trust, and open any link the AI gives you. If a source does not exist, it was made up. More in How AI works and Myths vs Facts. For medical, legal and money decisions, talk to a qualified person; these prompts are for learning and fun.