{"name":"PromptsBuddy prompt library","url":"https://promptsbuddy.com","license":"CC-BY-4.0","generatedFrom":"https://github.com/promptsbuddy/promptsbuddy/tree/main/content/prompts","count":38,"prompts":[{"slug":"one-pager-that-gets-forwarded","url":"https://promptsbuddy.com/prompts/one-pager-that-gets-forwarded","title":"A one-pager that gets forwarded","description":"The leave-behind PDF that has to work with nobody explaining it — written for the person your champion sends it to, not for your champion.","category":"Marketing","tags":["one-pager","pdf","collateral","b2b"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Lena Fischer","version":"1.0","published":"2026-08-29","updated":"2026-08-29","variables":[{"name":"OFFERING","description":"What you're describing — product, service, proposal or programme.","example":"A reconciliation tool for finance teams at 20–200 person companies."},{"name":"CHAMPION","description":"Who receives it first and what they want to happen.","example":"The finance manager who already likes us and needs budget approved."},{"name":"REAL_READER","description":"Who they forward it to — the person who decides and has no context.","example":"The CFO, who will read it for 40 seconds between meetings."},{"name":"PROOF","description":"Numbers, results, references you can actually stand behind.","example":"Cut month-end close from 6 days to 2 at one customer; 3 references available."}],"prompt":"Write a one-pager. It has to work without me in the room.\n\nWHAT IT'S ABOUT: {{OFFERING}}\nWHO GETS IT FIRST: {{CHAMPION}}\nWHO THEY FORWARD IT TO: {{REAL_READER}}\nPROOF I CAN STAND BEHIND:\n{{PROOF}}\n\nWrite it for the second reader — the one with no context and forty seconds.\n\nStructure, one page:\n- **A headline** stating the outcome, not the category. Not \"AI-powered\n  reconciliation\" — what changes for them.\n- **Three sentences** on the problem, in the second reader's language, using\n  their unit of pain (days, headcount, risk, money).\n- **How it works**, three steps maximum. No architecture.\n- **Proof**, using only what I gave you, with the number first.\n- **What it costs and what happens next**, stated plainly. Never \"contact us\n  for pricing\" without a range or a reason.\n- **One honest limitation.** Who this is not for.\n\nThen give me:\n- **The forty-second test.** What the second reader takes away if they read only\n  the headline and the bold text. Write it out.\n- **What my champion needs that this doesn't give them**, and whether that\n  belongs in a separate note rather than on the page.\n\nRules:\n- Do not use \"leading\", \"innovative\", \"seamless\", \"end-to-end\", or \"solution\".\n- Do not exceed one page. If it doesn't fit, cut the how-it-works section first.\n- Do not invent proof. Write [NEEDS PROOF] instead.\n- Do not write a call-to-action that requires a meeting as the only next step.","tips":["Writing for the forwarded-to reader is the whole trick. Your champion already believes you; the page isn't for them.","The honest limitation is what makes the rest credible, and it disqualifies bad-fit leads before they cost you a call.","Run the forty-second test on a colleague who knows nothing about it. If they get it wrong, the headline is the problem."]},{"slug":"pitch-deck-that-survives-slide-three","url":"https://promptsbuddy.com/prompts/pitch-deck-that-survives-slide-three","title":"A pitch deck that survives slide three","description":"Builds the narrative spine before any slide exists, then writes headlines that carry the argument on their own — so the deck still works when someone flicks through it without you.","category":"Marketing","tags":["presentation","slides","pitch-deck","storytelling"],"models":["claude","gpt","gemini"],"difficulty":"intermediate","author":"Rony","version":"1.0","published":"2026-08-29","updated":"2026-08-29","variables":[{"name":"PURPOSE","description":"What you need the audience to do when it ends. One action, not 'understand'.","example":"Approve £40k to run the pilot in Q1."},{"name":"AUDIENCE","description":"Who is in the room, what they already believe, and what they'll be sceptical of.","example":"CFO and two ops directors. They think last year's tooling spend didn't pay off."},{"name":"MATERIAL","description":"Everything you'd put in — numbers, findings, context. Dump it unsorted.","example":"Paste your notes, data, and the half-written slides you already have."}],"prompt":"Help me build a deck. Do not design slides yet.\n\nWHAT THEY SHOULD DO AFTERWARDS: {{PURPOSE}}\nWHO'S IN THE ROOM: {{AUDIENCE}}\nMY RAW MATERIAL:\n{{MATERIAL}}\n\n**1. The one sentence.** If they remember a single line, what is it? Write it\nas they would repeat it to someone who wasn't there.\n\n**2. The spine.** Six to ten beats that get from where the audience starts to\nthe action I want, each as a claim rather than a topic. \"Onboarding takes nine\ndays\" is a beat; \"Onboarding\" is not. Say which beat is the turn — the moment\nthe argument becomes undeniable.\n\n**3. The slides.** For each beat: a headline that states the claim in full, the\nsingle piece of evidence that supports it, and what goes on the slide visually.\nOne idea per slide. If a beat needs two slides, split it.\n\n**4. Cut list.** What in my material does not serve the spine. Be ruthless —\nname it and say why it goes.\n\n**5. The objection.** The hardest question this audience asks, where in the\ndeck it will occur to them, and whether to answer it before they ask.\n\nRules:\n- Headlines must carry the argument alone. A reader flicking through with no\n  presenter should get the whole case from headlines only.\n- Do not write \"Overview\", \"Background\", \"Our Solution\", or \"Thank you\" slides.\n- Do not put more than one number on a slide unless they are being compared.\n- Do not invent data. Mark anything I have not supplied as [NEEDS DATA].","tips":["The headline test is the whole thing: read only the headlines top to bottom. If it doesn't argue, the deck doesn't either.","Decks get forwarded far more than they get presented. Build for the reader, not the room.","Section 4 hurts and is where the deck gets good. Most decks are 40% material the author was attached to."]},{"slug":"social-posts-from-one-real-thing","url":"https://promptsbuddy.com/prompts/social-posts-from-one-real-thing","title":"Social posts from one real thing that happened","description":"Turns a single concrete event — a customer call, a shipped fix, a number that surprised you — into posts worth reading, instead of generating content about nothing.","category":"Marketing","tags":["social-media","content","marketing","writing"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Rony","version":"1.0","published":"2026-08-29","updated":"2026-08-29","variables":[{"name":"WHAT_HAPPENED","description":"One specific real thing. A conversation, a result, a mistake, a number.","example":"A customer told us they'd been doing month-end reconciliation by hand for four years because nobody had asked."},{"name":"CHANNEL","description":"Where it's going — the format constraints differ a lot.","example":"LinkedIn, where our audience is finance and ops managers."},{"name":"POV","description":"Whose voice, and what they're known for.","example":"Our founder, who writes plainly and dislikes hustle language."}],"prompt":"Here is one real thing that happened. Turn it into social posts.\n\nWHAT HAPPENED:\n{{WHAT_HAPPENED}}\n\nCHANNEL: {{CHANNEL}}\nWHOSE VOICE: {{POV}}\n\nFirst, tell me: **is there actually something here?** If this is only\ninteresting to us and not to a reader, say so and stop. Do not manufacture\nsignificance.\n\nIf there is, write **three posts, three different angles**:\n- The observation — what this reveals that most people get wrong.\n- The specific — the story told plainly, with the detail that makes it real.\n- The useful — what a reader can do differently tomorrow because of it.\n\nFor each: the opening line separately, since that is the only part most people\nread. Keep the concrete detail; it is the whole reason the post works.\n\nThen flag:\n- **Which of the three is weakest**, and why.\n- **Anything I should check before posting** — a claim about a customer, a\n  number I'd need permission to share, a person identifiable from the detail.\n\nRules:\n- Do not open with a rhetorical question, \"Let that sink in\", or a one-word\n  line followed by a full stop.\n- No engagement bait, no \"agree?\", no thread-teasing.\n- Do not use an em dash more than once per post.\n- Do not generalise the specific detail away. \"A customer\" is worse than \"a\n  finance manager at a 40-person logistics firm\".\n- If the story involves a real customer, assume I need their permission and say so.","tips":["The gate at the top is the point — most social content is generated from nothing, and it shows.","One real event beats ten content ideas. Keep a note of things that surprise you; that note is your content calendar.","The permission check has saved people from posting a customer story they hadn't cleared."]},{"slug":"decide-between-two-options","url":"https://promptsbuddy.com/prompts/decide-between-two-options","title":"Decide between two options you keep going back and forth on","description":"Surfaces the criterion you're actually optimising for, tests each option against being wrong, and names the information that would settle it — or tells you it's already a coin flip.","category":"Strategy","tags":["decisions","thinking","choices","clarity"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Rony","version":"1.0","published":"2026-08-28","updated":"2026-08-28","variables":[{"name":"OPTION_A","description":"The first option, and what draws you to it.","example":"Take the bigger job — more money, more scope, a manager I don't know."},{"name":"OPTION_B","description":"The second, and what draws you to that.","example":"Stay — work I like, people I trust, no obvious next step."},{"name":"STUCK","description":"Why you haven't decided. This is usually the most revealing field.","example":"Every time I pick one I immediately think about what I'd lose from the other."}],"prompt":"I keep going back and forth between two options.\n\nOPTION A: {{OPTION_A}}\nOPTION B: {{OPTION_B}}\nWHY I'M STUCK: {{STUCK}}\n\n**1. What am I actually optimising for?** From how I described these, infer\nthe criterion I am really using — not the one I would say out loud. Name it in\none sentence. If I appear to be optimising for two things that conflict, say so;\nthat is usually the whole problem.\n\n**2. The reversibility test.** For each option: if it turns out wrong in a\nyear, what does undoing it cost? Cheap-to-reverse decisions deserve less\nagonising than I am giving them.\n\n**3. What I lose either way.** Name the specific thing each option forecloses.\nDo not soften this — the fear of loss is what has me stuck, and seeing it\nwritten down usually shrinks it.\n\n**4. The one fact that would settle this.** Something I could find out this\nweek. If there genuinely isn't one, say so.\n\n**5. Your call.** Pick one and give the single strongest reason. If the honest\nanswer is that these are close enough that the choice matters less than\ncommitting, say exactly that.\n\nRules:\n- Do not produce a weighted scoring table. I will game the weights to get the\n  answer I already want.\n- Do not tell me to trust my gut.\n- Do not hedge in section 5. Pick one.\n- Work only from what I told you. Ask nothing.","tips":["Section 1 ends most of these. 'You're optimising for not feeling foolish' lands harder than any pros and cons list.","The reversibility test is the practical one — most decisions people agonise over are cheap to undo, and most they rush are not.","If it says the options are close enough that committing matters more, believe it. That's a real finding, not a dodge."]},{"slug":"explain-it-to-someone-who-isnt-an-expert","url":"https://promptsbuddy.com/prompts/explain-it-to-someone-who-isnt-an-expert","title":"Explain it to someone who isn't an expert","description":"Rewrites a technical explanation for a specific real person, keeps every fact intact, and flags where the simplification starts to become wrong.","category":"Communication","tags":["explaining","communication","teaching","clarity"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Meghna","version":"1.0","published":"2026-08-28","updated":"2026-08-28","variables":[{"name":"EXPLANATION","description":"What you'd say to a peer — jargon and all.","example":"Paste your technical explanation, spec, or diagnosis."},{"name":"READER","description":"One specific person, not a category. What they already know matters more than their job title.","example":"My client's finance director — comfortable with spreadsheets, has never heard of an API."},{"name":"WHY","description":"What they need to do with it. This decides what can be left out.","example":"Decide whether to approve three more weeks of work."}],"prompt":"Rewrite this for someone who is not an expert.\n\nWHAT I'D SAY TO A PEER:\n{{EXPLANATION}}\n\nWHO'S READING: {{READER}}\nWHAT THEY NEED TO DO WITH IT: {{WHY}}\n\nGive me:\n\n**1. The version for them.** Same facts, no jargon that isn't defined on\nfirst use, and ordered by what they need for their decision rather than by how\nthe system works. Lead with the part that affects them.\n\n**2. What I can leave out.** The details that are true, that I am attached to,\nand that do not change what they will do. Being explicit about this is what\nkeeps the explanation short.\n\n**3. Where simplification becomes wrong.** Any place the plain-language version\nis now slightly inaccurate, and whether that matters for this reader's decision.\nThis is the important section.\n\n**4. The question they'll ask.** The most likely follow-up, and a one-sentence\nanswer ready.\n\nRules:\n- Do not use an analogy unless the mechanism genuinely matches. A wrong analogy\n  is worse than jargon because it feels understood.\n- Do not remove numbers, caveats, or uncertainty. Simplify the language, not the\n  facts.\n- Do not condescend. Not an expert is not the same as not intelligent.\n- If the explanation cannot be made honest at this reader's level, say so and\n  say what they would need to know first.","tips":["Name one real person in READER. 'Non-technical stakeholder' produces a generic rewrite; 'my client's finance director' produces a usable one.","Section 3 is what makes this safe to send — you know exactly which corner you cut.","Works on your own writing before a presentation, and on someone else's before you have to relay it."]},{"slug":"prepare-for-a-hard-conversation","url":"https://promptsbuddy.com/prompts/prepare-for-a-hard-conversation","title":"Prepare for a conversation you've been avoiding","description":"Works out what you actually want from it, what the other person is likely protecting, and the first two sentences — the part everyone rehearses badly.","category":"Communication","tags":["communication","conflict","difficult-conversations","preparation"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Dinesh","version":"1.0","published":"2026-08-28","updated":"2026-08-28","variables":[{"name":"SITUATION","description":"What's happened, in your own unfiltered words. Vent if you need to.","example":"My landlord hasn't fixed the heating in six weeks and stopped replying."},{"name":"THEM","description":"Who they are to you, and what you know about their position.","example":"Manages four buildings, always says he's short-staffed, holds my deposit."},{"name":"OUTCOME","description":"What you want to be true afterwards.","example":"Heating fixed within a week, and a written record in case it isn't."}],"prompt":"Help me prepare for a conversation I have been putting off.\n\nTHE SITUATION:\n{{SITUATION}}\n\nTHE OTHER PERSON: {{THEM}}\nWHAT I WANT AFTERWARDS: {{OUTCOME}}\n\nWork through:\n\n**1. What I actually want.** Separate the outcome I stated from what I also\nwant emotionally — to be right, to be apologised to, to stop feeling anxious.\nName both plainly. They often conflict, and knowing which one I'm optimising\nfor changes what I should say.\n\n**2. Their side, argued properly.** What does this look like from where they\nsit? What are they protecting — money, time, reputation, their own boss? Argue\nit as they would, not as a strawman.\n\n**3. My opening.** The first two sentences, written out. Specific, calm, no\npreamble, and stating the outcome I want rather than relitigating history.\n\n**4. The three things they might say**, and a response to each that neither\ncaves nor escalates. Include what to do if they get defensive.\n\n**5. My floor.** The minimum I will accept, and what I do if I don't get it.\nDecide this before the conversation, not during.\n\nRules:\n- Do not tell me to stay calm or use \"I feel\" statements. Give me actual words.\n- Do not assume the other person is acting in bad faith unless I said so.\n- Do not soften my ask to the point where it stops being an ask.\n- If what I want is unreasonable given what I have told you, say so.","tips":["Section 1 is the one that changes outcomes. Half of these conversations go badly because you wanted an apology and asked for a repair.","Deciding your floor in advance is what stops you accepting something in the moment that you'll resent for months.","Read your opening out loud. If it sounds like a script, cut it down until it sounds like you."]},{"slug":"say-no-and-keep-the-relationship","url":"https://promptsbuddy.com/prompts/say-no-and-keep-the-relationship","title":"Say no and keep the relationship","description":"Declines a request clearly, without the apologising, over-explaining, or fake availability that turns a no into a maybe and gets you asked again next week.","category":"Communication","tags":["communication","boundaries","email","work"],"models":["claude","gpt","gemini","grok"],"difficulty":"beginner","author":"Rony","version":"1.0","published":"2026-08-28","updated":"2026-08-28","variables":[{"name":"REQUEST","description":"What was asked, pasted or described.","example":"Can you join the Thursday steering committee? It's weekly, 90 minutes."},{"name":"WHY_NOT","description":"The real reason, even if you wouldn't say it out loud.","example":"It's not my area, and I'd be there to make someone else's meeting look important."},{"name":"RELATIONSHIP","description":"Who's asking and what you need the relationship to survive as.","example":"A peer in another team I'll need favours from next quarter."}],"prompt":"Help me say no.\n\nTHE REQUEST:\n{{REQUEST}}\n\nTHE REAL REASON:\n{{WHY_NOT}}\n\nWHO'S ASKING: {{RELATIONSHIP}}\n\nFirst, tell me honestly: **should this be a no?** Given what I've said, is\ndeclining the right call, or am I avoiding something I should just do? One\nshort paragraph. If it should be a yes, say so and stop.\n\nIf it's a no, write the reply:\n- Decline in the first sentence. Not the third.\n- One reason, not three. Stacking reasons sounds like an excuse and invites\n  negotiation on each one.\n- Do not apologise more than once, and not at all if I have done nothing wrong.\n- Offer an alternative only if I would genuinely do it. Never invent\n  availability I do not have.\n- End without leaving the door ajar. No \"maybe next quarter\" unless I mean it.\n\nThen flag: **what could make them ask again?** Any phrasing that reads as a\nsoft no rather than a no.\n\nKeep it under 80 words and match how I would actually talk to this person.","tips":["The honesty check catches the times you're avoiding, not declining. It says yes more often than you'd expect.","One reason. Three reasons is a negotiation surface.","The 'could they ask again' flag is what stops the same request returning every month."]},{"slug":"summarise-a-document-you-have-to-act-on","url":"https://promptsbuddy.com/prompts/summarise-a-document-you-have-to-act-on","title":"Summarise a document you actually have to act on","description":"For a contract, policy, or report where getting it wrong costs you. Separates what it obliges you to do from what it merely says, and flags the clauses worth a second read.","category":"Research","tags":["reading","documents","contracts","comprehension"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Lena Fischer","version":"1.0","published":"2026-08-28","updated":"2026-08-28","variables":[{"name":"DOCUMENT","description":"The document. Paste the whole thing if you can — partial text produces partial answers.","example":"A lease, an MSA, an insurance policy, a 40-page report."},{"name":"MY_POSITION","description":"Which side you're on and what you're deciding.","example":"I'm the tenant, deciding whether to sign a two-year term."}],"prompt":"I have to act on this document, not just understand it.\n\nDOCUMENT:\n{{DOCUMENT}}\n\nMY POSITION: {{MY_POSITION}}\n\nGive me:\n\n**1. What this obliges me to do.** Only actual obligations on my side —\ndeadlines, payments, notice periods, things I must not do. Quote the clause\nfor each. Do not include obligations on the other party here.\n\n**2. What it lets the other side do.** Especially anything they can do\nunilaterally: change terms, terminate, increase a price, keep something of mine.\n\n**3. The three clauses I should read again myself.** The ones where the wording\nis doing more work than it appears to. Say what to watch for in each.\n\n**4. What is missing.** Things a document of this kind normally covers that\nthis one does not. Absence is often the risk.\n\n**5. Plain-language summary.** Five sentences, no jargon, as if explaining to\nsomeone who has not read it.\n\nRules:\n- Do not give legal, medical, or financial advice. Describe what the document\n  says and what it leaves unsaid; the decision is mine.\n- Do not paraphrase a clause in a way that softens it. Quote when the exact\n  wording matters.\n- If a section is ambiguous, say it is ambiguous rather than picking a reading.\n- If I have given you only part of the document, say what you cannot assess.","tips":["Section 4 catches the most: no cap on liability, no exit clause, no defined response time.","Always read the three flagged clauses yourself. This narrows where to look — it doesn't replace looking.","For anything with real money or legal exposure, this is preparation for talking to a professional, not a substitute for one."]},{"slug":"triage-a-week-you-cant-finish","url":"https://promptsbuddy.com/prompts/triage-a-week-you-cant-finish","title":"Triage a week you can't possibly finish","description":"Takes an overloaded list and sorts it into what actually ships, what gets dropped on purpose, and what someone needs to be told — instead of pretending it all fits.","category":"Productivity","tags":["productivity","planning","prioritisation","workload"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Sam Reyes","version":"1.0","published":"2026-08-28","updated":"2026-08-28","variables":[{"name":"LIST","description":"Everything on your plate, dumped in any order.","example":"Paste your todo list, inbox flags, and the things you're carrying in your head."},{"name":"TIME","description":"Honestly available hours, after meetings and interruptions.","example":"Maybe 11 focused hours across the week; Tuesday is entirely meetings."},{"name":"CONSEQUENCES","description":"Anything with a hard deadline or a person waiting on it.","example":"The board deck is Thursday. Two people are blocked on my review."}],"prompt":"This does not all fit. Help me choose rather than pretend.\n\nEVERYTHING ON MY LIST:\n{{LIST}}\n\nTIME I ACTUALLY HAVE: {{TIME}}\nHARD DEADLINES AND PEOPLE WAITING: {{CONSEQUENCES}}\n\nSort everything into exactly four groups:\n\n**Ships this week.** What fits in the hours I gave you, in the order I should\ndo it. Include a rough time estimate per item and total it — if the total\nexceeds my hours, cut until it doesn't.\n\n**Unblocks someone else.** Anything where a person is waiting on me. These\nusually cost me minutes and cost them days, so they go early even if they feel\nunimportant.\n\n**Dropped on purpose.** Things I am consciously not doing. Not \"later\" —\ndropped. For each, one line on what happens as a result.\n\n**Someone needs to know.** Which dropped or delayed items require me to tell a\nspecific person, and the one-sentence message.\n\nRules:\n- Do not create a fifth \"if there's time\" bucket. There is no time.\n- Do not tell me to wake up earlier, timebox, or use a productivity system.\n- If something on my list is not actually mine to do, say so.\n- If the hard deadlines alone exceed my hours, lead with that.","tips":["'Unblocks someone else' is the group people systematically under-rate. Ten minutes of yours can be two days of theirs.","The fourth group is the one that saves you. A dropped task nobody was told about is how trust goes.","Be honest about TIME. Padding it just moves the failure to Friday."]},{"slug":"read-a-paper-without-losing-the-caveats","url":"https://promptsbuddy.com/prompts/read-a-paper-without-losing-the-caveats","title":"Read a paper without losing the caveats","description":"Extracts what a study actually established, at what sample size, under what conditions — and separates the authors' claims from what the press release said.","category":"Research","tags":["research","papers","critical-reading","evidence"],"models":["claude","gpt","perplexity"],"difficulty":"intermediate","author":"Lena Fischer","version":"1.0","published":"2026-08-27","updated":"2026-08-27","variables":[{"name":"PAPER","description":"The abstract and methods at minimum; the full text if you have it.","example":"Paste the paper or its abstract, methods and results."},{"name":"WHY","description":"What decision you're trying to inform — this changes what matters.","example":"Deciding whether to switch our onboarding to the pattern this study tested."}],"prompt":"I want to understand what this paper actually shows.\n\nPAPER:\n{{PAPER}}\n\nWHAT I'M TRYING TO DECIDE: {{WHY}}\n\n**1. The claim.** What the authors say they found, in one sentence, in their\nown terms.\n\n**2. What was actually measured.** The operational definitions — what exact\nthing was counted or scored, on whom, how many, over what period, compared\nagainst what. Numbers, not descriptions.\n\n**3. The gap.** The distance between what was measured and what the claim\nimplies. This is the section that matters.\n\n**4. Conditions.** What had to be true for this result to hold, including\npopulation, setting, and anything the authors themselves flag in limitations.\nQuote their limitations rather than paraphrasing.\n\n**5. Does it bear on my decision?** Given what I said I am deciding, state\nplainly whether this study is evidence for it, weak evidence, or not evidence\nat all. Say which of my conditions differ from the study's.\n\nRules:\n- Do not report a p-value as if it were an effect size. Give the effect size\n  and its interval if present; say so if absent.\n- Do not smooth over a small sample. Name n.\n- Do not use the word \"proves\".\n- If the methods section is not in what I gave you, say what you cannot assess\n  rather than assuming it was done well.","tips":["Section 3 is the whole prompt. Most misreading happens in the gap between the measurement and the headline.","Give it your actual decision — the same paper is strong evidence for one question and irrelevant to another.","Quote-the-limitations beats paraphrase: authors are usually honest in that section and it gets lost in summaries."]},{"slug":"debug-from-a-stack-trace","url":"https://promptsbuddy.com/prompts/debug-from-a-stack-trace","title":"Debug from a stack trace without guessing","description":"Forces the model to separate what the trace proves from what it's inferring, rank hypotheses by testability, and give you the cheapest experiment that discriminates between them.","category":"Engineering","tags":["debugging","engineering","troubleshooting","coding"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Meghna","version":"1.0","published":"2026-08-26","updated":"2026-08-26","variables":[{"name":"TRACE","description":"The full stack trace or error output. Don't trim it.","example":"Paste the whole trace, including the framework frames."},{"name":"CODE","description":"The function at the top of your own frames, plus anything it calls.","example":"Paste the relevant source."},{"name":"WHEN","description":"When it happens and when it doesn't — the most useful thing you can supply.","example":"Only in production, only for users with more than 50 saved items, started Tuesday."}],"prompt":"Help me debug this. Do not propose a fix yet.\n\nTRACE:\n{{TRACE}}\n\nCODE:\n{{CODE}}\n\nWHEN IT HAPPENS: {{WHEN}}\n\n**1. What the trace proves.** State only what is certain from the trace\nitself — which call failed, with what, in which frame. Separate this from\nanything you are inferring. Label inferences `[inferred]`.\n\n**2. What \"when it happens\" rules out.** Given the conditions I described,\nwhich otherwise-plausible causes are eliminated? This is the most useful\nsection — be rigorous.\n\n**3. Hypotheses, ranked by testability**, not by likelihood. For each: the\nmechanism in one sentence, and the single cheapest observation that would\nconfirm or kill it. Prefer a log line or a one-off query over a code change.\n\n**4. The discriminating test.** If you could run exactly one thing, what is it,\nand what does each possible outcome tell me?\n\nRules:\n- Do not suggest a fix until I come back with results.\n- Do not suggest \"add error handling\" or \"add logging\" as a hypothesis. Those\n  are not causes.\n- If the trace is truncated or the relevant frame is missing, say exactly what\n  else you need rather than working around it.\n- Do not assume library internals you have not seen. Say when you are.","tips":["The WHEN field does the heavy lifting. 'Only for users with >50 items' eliminates more hypotheses than the trace does.","Ranking by testability rather than likelihood is deliberate — a cheap test for an unlikely cause beats an expensive test for a likely one.","Come back with the test result and only then ask for a fix. Fixes proposed before evidence are how you end up with three unrelated changes."]},{"slug":"performance-feedback-that-lands","url":"https://promptsbuddy.com/prompts/performance-feedback-that-lands","title":"Performance feedback that lands instead of stinging","description":"Turns a vague concern about someone's work into specific, dated, behavioural feedback — and checks whether what you're about to say is actually about them or about your own preferences.","category":"Management","tags":["management","feedback","people","communication"],"models":["claude","gpt"],"difficulty":"advanced","author":"Dinesh","version":"1.0","published":"2026-08-26","updated":"2026-08-26","variables":[{"name":"CONCERN","description":"What's bothering you, in your own unfiltered words.","example":"They over-engineer everything and it's slowing the team down."},{"name":"EVIDENCE","description":"Specific instances with dates. If you can't list three, that's a finding.","example":"The caching layer in March; the config refactor in May; the abstraction on the export job last week."},{"name":"CONTEXT","description":"Their level, how long they've been here, what they're working toward.","example":"Mid-level, 18 months in, wants to be promoted to senior this cycle."}],"prompt":"Help me give this feedback well.\n\nMY CONCERN: {{CONCERN}}\nSPECIFIC INSTANCES: {{EVIDENCE}}\nABOUT THEM: {{CONTEXT}}\n\n**1. Pressure-test me first.** Before drafting anything, answer:\n- Is this a pattern or an incident? The evidence I gave — does it support a pattern?\n- Is this about impact on the work, or about a preference of mine? Be blunt.\n- What might be causing it that I have not considered, including something I\n  am doing?\nIf the evidence does not support the concern, say so and stop. Do not draft.\n\n**2. The behavioural version.** Restate the concern as observable behaviour and\nits effect — no adjectives about the person, no \"you always\".\n\n**3. The conversation.** What I actually say, in order: the specific opening,\nthe evidence, the space for their side, the concrete change I am asking for,\nand how we will both know it changed. Keep my words plain.\n\n**4. Their likely response**, and how I respond to that without retreating or\nescalating.\n\nRules:\n- Do not use a feedback sandwich.\n- Do not characterise the person. Behaviour and effect only.\n- Do not soften the ask to the point where they could leave unsure what to change.\n- Do not invent evidence. Work only from what I gave you.","tips":["Step 1 stops perhaps a third of these conversations from happening, which is the right outcome.","'Is this impact or preference?' is uncomfortable and is the question worth paying for.","If you can't list three dated instances, you have an impression, not feedback. Wait."]},{"slug":"read-a-spreadsheet-like-an-analyst","url":"https://promptsbuddy.com/prompts/read-a-spreadsheet-like-an-analyst","title":"Read a spreadsheet like an analyst, not a summariser","description":"Interrogates a dataset before describing it: what's missing, what's double-counted, which column is lying, and only then what it appears to say.","category":"Data","tags":["data","analysis","spreadsheet","quality"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Sam Reyes","version":"1.0","published":"2026-08-26","updated":"2026-08-26","variables":[{"name":"DATA","description":"Paste the sheet as CSV, or the first 50 rows plus the column list.","example":"Paste CSV, headers included."},{"name":"QUESTION","description":"What you're hoping it tells you.","example":"Whether churn got worse after the March pricing change."},{"name":"PROVENANCE","description":"Where the data came from and how it was collected — this is where most errors live.","example":"Exported from Stripe, joined to a hand-maintained CRM sheet by email."}],"prompt":"Here is a dataset and the question I want answered.\n\nDATA:\n{{DATA}}\n\nQUESTION: {{QUESTION}}\nWHERE IT CAME FROM: {{PROVENANCE}}\n\nDo not answer the question yet. Work in this order:\n\n**1. What this data can and cannot answer.** Given the columns and the\nprovenance, state plainly whether my question is answerable at all. If it is\nonly partly answerable, say which part.\n\n**2. Integrity check.** Look for: missing values and whether they're missing\nat random, duplicate rows, a join that could double-count, dates in mixed\nformats or timezones, outliers that look like data-entry errors, and columns\nwhose meaning is ambiguous from the header alone. Quote example rows.\n\n**3. The one column I'd distrust most**, and why.\n\n**4. Now answer**, with the specific caveat that belongs next to the number.\n\n**5. What would make this trustworthy** — the extra column, source, or check\nthat would move this from suggestive to solid.\n\nRules:\n- Do not compute a summary statistic before step 2. A mean over dirty data is\n  worse than no number.\n- Do not infer causation from a before/after split. Say what else changed in\n  that window that you would need to rule out.\n- If the sample is too small to support the question, say the number and stop.","tips":["Provenance is the highest-value field. Most data errors are joining errors, and the model can't see the join from the CSV.","Step 3 is uncannily good — 'the column I'd distrust most' catches things like a status field that was backfilled.","Never paste customer data. Anonymise, or use a structurally identical sample."]},{"slug":"status-update-that-survives-skimming","url":"https://promptsbuddy.com/prompts/status-update-that-survives-skimming","title":"A status update that survives being skimmed","description":"Front-loads the one thing that changed, states what's at risk without hedging, and makes the ask explicit — so the person who reads only the first line still gets it.","category":"Product","tags":["communication","async","status","writing"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Rony","version":"1.0","published":"2026-08-26","updated":"2026-08-26","variables":[{"name":"RAW","description":"What happened, in whatever order it comes out. Tickets, notes, Slack.","example":"Paste your notes for the week."},{"name":"AUDIENCE","description":"Who reads it and what they'll do with it.","example":"VP Eng, who'll skim on a phone between meetings and only act if something's blocked."}],"prompt":"Turn this into a status update for {{AUDIENCE}}.\n\nRAW:\n{{RAW}}\n\nFormat, in this order:\n\n**Line one.** The single most important change since last time. If someone\nreads nothing else, this is what they need. No preamble.\n\n**Status.** On track / at risk / off track. Pick one. If at risk or off track,\nthe next sentence is the reason, not the mitigation.\n\n**What moved.** Three bullets maximum, each an outcome rather than an activity.\n\"Checkout flow is live for 10% of traffic\", not \"worked on checkout\".\n\n**What I need.** The specific ask, with a name and a date, or the words\n\"nothing right now\". Never leave this section vague.\n\n**What I'm watching.** One risk, and the signal that would tell us it's real.\n\nRules:\n- Do not hedge the status. \"Mostly on track\" is off track.\n- Do not list activity. If it produced no outcome, leave it out.\n- Do not bury a blocker in the third bullet. It goes in line one.\n- Keep the whole thing under 150 words.","tips":["'The reason, not the mitigation' is the rule people break. Leading with the fix hides how bad it is.","Keep the format identical week to week — the diff between updates becomes the real signal.","If 'what I need' is always 'nothing', either you're stuck and not saying so, or you don't need the update."]},{"slug":"the-scope-cut","url":"https://promptsbuddy.com/prompts/the-scope-cut","title":"The scope cut that still ships something real","description":"For when the date won't move. Sorts the plan into what ships, what waits, and what quietly gets cancelled — and names what breaks for users in each case.","category":"Product","tags":["product","planning","scope","project-management"],"models":["claude","gpt","gemini"],"difficulty":"intermediate","author":"Lena Fischer","version":"1.0","published":"2026-08-26","updated":"2026-08-26","variables":[{"name":"SCOPE","description":"Everything currently in the plan. A bulleted dump is fine.","example":"Paste the epic list or the PRD's requirements section."},{"name":"DEADLINE","description":"The fixed date and what makes it fixed.","example":"March 14 — it's tied to a conference keynote, so it genuinely cannot move."},{"name":"CAPACITY","description":"Who is actually available, honestly, including partial allocations.","example":"Two engineers full time, one designer half time, and me at maybe 30%."}],"prompt":"The date is fixed and the scope is not going to fit.\n\nSCOPE:\n{{SCOPE}}\n\nDEADLINE: {{DEADLINE}}\nWHO'S ACTUALLY AVAILABLE: {{CAPACITY}}\n\nProduce:\n\n**1. The spine.** The smallest set of items that still delivers a coherent\nexperience — something a user could complete end to end without hitting a dead\nend. Not the highest-value items; the ones that make the others make sense.\n\n**2. Three buckets.** Sort everything else into: *ships if it goes well*,\n*explicitly deferred*, and *quietly cancelled*. Be willing to put things in the\nthird bucket — a plan where nothing is cancelled is not a cut.\n\n**3. What breaks.** For each deferred and cancelled item, one sentence on what\na user cannot do, and who will notice first.\n\n**4. The conversation.** What I have to tell each stakeholder, in their terms,\nincluding the one who will be angriest and why they are not wrong.\n\n**5. The tell.** The specific signal, two weeks out, that even the spine is\nnot going to make it.\n\nRules:\n- Do not propose adding people. Do not propose moving the date.\n- Do not suggest \"reducing polish\" as a cut. Name the actual thing that goes.\n- If the spine still does not fit the capacity I gave you, say so directly and\n  say what a smaller coherent product would be.","tips":["'Quietly cancelled' is the bucket that matters. Deferring everything is how backlogs become fiction.","Section 4 is what makes this shippable politically — a cut nobody has been told about isn't a cut.","Run it before you're behind, not after. It reads as planning in January and as panic in March."]},{"slug":"the-test-you-would-have-forgotten","url":"https://promptsbuddy.com/prompts/the-test-you-would-have-forgotten","title":"The test case you would have forgotten","description":"Reads a function and enumerates the boundaries, empty states, and concurrency cases you didn't think of — then writes only the tests that could actually fail.","category":"Engineering","tags":["testing","engineering","quality","coding"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Meghna","version":"1.0","published":"2026-08-26","updated":"2026-08-26","variables":[{"name":"CODE","description":"The function or module under test, plus its types.","example":"Paste the source."},{"name":"STACK","description":"Language and test framework, so the output is runnable.","example":"TypeScript, Vitest, no mocking library."},{"name":"EXISTING","description":"Tests that already exist, so it doesn't duplicate them. 'None' is fine.","example":"One happy-path test for the basic case."}],"prompt":"Find the cases I have not thought about.\n\nCODE:\n{{CODE}}\n\nSTACK: {{STACK}}\nTESTS THAT ALREADY EXIST: {{EXISTING}}\n\n**1. Enumerate the input space.** For each parameter: the boundary values, the\nempty or zero case, the maximum, the malformed case, and anything nullable.\nThen the combinations that interact.\n\n**2. Beyond inputs.** Ordering and concurrency, partial failure, repeated calls\n(is this idempotent?), timezone and locale, and what happens if a dependency is\nslow rather than down.\n\n**3. Rank by \"could this actually happen and would it hurt?\"** Drop anything\nthat is theoretically possible but harmless — say you dropped it and why.\n\n**4. Write the top tests** for {{STACK}}. Each test name states the behaviour,\nnot the method name. Include the assertion that would fail today if the bug\nwere present.\n\nRules:\n- Do not write tests that restate the implementation. If a test would pass for\n  any implementation of this signature, it is worthless — skip it.\n- Do not duplicate what already exists.\n- Do not mock the thing under test.\n- If a case cannot be tested without changing the code's shape, say so and say\n  what change would make it testable.","tips":["Section 1 is worth reading even if you write the tests yourself — the enumeration is the part humans skip.","The 'would fail today' requirement filters out tests that assert nothing.","Ask about idempotency explicitly for anything that touches money, email, or external APIs."]},{"slug":"cold-email-that-earns-a-reply","url":"https://promptsbuddy.com/prompts/cold-email-that-earns-a-reply","title":"A cold email that earns a reply","description":"Under 90 words, one specific observation about them, one ask small enough to say yes to. Includes the test for whether it's actually personalised or just mail-merged.","category":"Sales","tags":["sales","email","outreach","writing"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Dinesh","version":"1.0","published":"2026-08-25","updated":"2026-08-25","variables":[{"name":"RECIPIENT","description":"Who they are and what you actually know about them or their company.","example":"Head of Ops at a 60-person logistics firm; they posted about switching WMS last month."},{"name":"OFFER","description":"What you do, in one plain sentence.","example":"We cut manual reconciliation between order and billing systems."},{"name":"ASK","description":"The single next step. Smaller is better.","example":"15 minutes, or just reply 'send it' and I'll email a two-minute demo video."}],"prompt":"Write a cold email. Hard limit: 90 words, excluding the signature.\n\nRECIPIENT: {{RECIPIENT}}\nWHAT WE DO: {{OFFER}}\nTHE ASK: {{ASK}}\n\nStructure:\n- Subject line: under 6 words, lowercase, no colons, and specific enough that\n  it could not be sent to a different company.\n- First line: something true about *them*. Not a compliment — an observation.\n- Second: the problem that observation implies, phrased as a question they\n  would recognise as their own.\n- Third: what we do, in one sentence, with a number if I gave you one.\n- Last: the ask, exactly as I framed it.\n\nThen run two checks and show both:\n1. **Mail-merge test.** Which sentences would still make sense sent to a\n   different company? List them — they need rewriting or cutting.\n2. **The reply cost.** How many minutes of work am I asking for? If the honest\n   answer is more than five, the ask is too big; say so.\n\nRules:\n- Do not open with \"I hope this finds you well\" or \"I'll be brief\".\n- Do not use \"quick question\", \"circling back\", \"touching base\", or \"synergy\".\n- Do not claim to have read something I did not tell you I read.\n- No paragraph over two sentences.","tips":["The mail-merge test kills the sentences that make cold email feel like cold email.","Shrink the ask until it's embarrassing. 'Reply with a word' converts far better than 'find 30 minutes'.","90 words is not a style choice — it's what fits in a phone preview before the fold."]},{"slug":"kill-the-flattery","url":"https://promptsbuddy.com/prompts/kill-the-flattery","title":"Kill the flattery","description":"A standing instruction that removes the praise, the restating, and the closing summary — so the reply is only the part you actually needed.","category":"Fundamentals","tags":["prompting","tone","short","fundamentals"],"models":["claude","gpt","gemini","grok"],"difficulty":"beginner","author":"Lena Fischer","version":"1.0","published":"2026-08-25","updated":"2026-08-25","variables":[{"name":"REQUEST","description":"Whatever you were going to ask anyway.","example":"Review this migration plan."}],"prompt":"For this conversation:\n\n- Do not open by praising the question, the idea, or me.\n- Do not restate my question before answering it.\n- Do not end with a summary of what you just said.\n- Do not offer to help further unless I ask.\n- If something is good, say so once, specifically, and move on.\n- If you disagree, lead with the disagreement rather than burying it after\n  two paragraphs of agreement.\n- Say \"I don't know\" when you don't know, instead of producing a confident\n  shape.\n\n{{REQUEST}}","tips":["Structural bans work; attitudinal ones don't. 'No closing summary' is enforceable, 'be more critical' is not.","The 'lead with the disagreement' line changes the most. Buried objections get skimmed past.","Paste it into your project or custom instructions once rather than per-conversation."]},{"slug":"lesson-plan-for-a-mixed-class","url":"https://promptsbuddy.com/prompts/lesson-plan-for-a-mixed-class","title":"A lesson plan for a class that isn't at one level","description":"Plans one lesson three ways at once — for the students who are behind, on track, and ahead — without turning it into three lessons you have to run simultaneously.","category":"Teaching","tags":["teaching","education","lesson-plan","classroom"],"models":["claude","gpt","gemini"],"difficulty":"intermediate","author":"Meghna","version":"1.0","published":"2026-08-25","updated":"2026-08-25","variables":[{"name":"TOPIC","description":"What you're teaching, and to whom.","example":"Photosynthesis, Year 8, 50 minutes."},{"name":"SPREAD","description":"The honest range in the room.","example":"Six students still shaky on cell structure; most fine; four already read ahead."},{"name":"CONSTRAINTS","description":"Time, materials, room, anything fixed.","example":"No lab access. One projector. Must include a written assessment."}],"prompt":"Plan one lesson.\n\nTOPIC: {{TOPIC}}\nTHE RANGE IN THE ROOM: {{SPREAD}}\nCONSTRAINTS: {{CONSTRAINTS}}\n\nProduce:\n\n**1. The one idea.** If they remember a single thing in a year, what is it?\nOne sentence, in student language.\n\n**2. The misconception.** The wrong belief students at this level reliably\narrive with, and the specific moment in the lesson that confronts it.\n\n**3. The spine.** A minute-by-minute run of the lesson that works for the\nmiddle of the room. Include what I am doing and what they are doing.\n\n**4. Two adjustments, not two lessons.** One thing to change for the students\nwho are behind and one for those ahead — each an adjustment to the *same*\nactivity, not a separate task I would have to supervise in parallel.\n\n**5. The check.** A question I can ask in the last four minutes that tells me\nwho actually got it. It must be answerable in one sentence and must not be\nanswerable correctly by someone who only half-followed.\n\nRules:\n- Do not produce a worksheet. Produce a plan.\n- Do not assume equipment beyond the constraints I gave.\n- Do not use \"differentiation\" or \"scaffolding\" as if naming them explains them\n  — describe the actual adjustment.\n- If 50 minutes is not enough for this topic, say what to cut.","tips":["Section 4's constraint — adjustments, not parallel tasks — is what makes this survive contact with a real classroom.","The exit check is the most reusable output. Build a bank of them.","Give it the honest spread. 'Mixed ability' produces a generic plan; 'six shaky on cell structure' produces a usable one."]},{"slug":"make-it-ask-before-it-answers","url":"https://promptsbuddy.com/prompts/make-it-ask-before-it-answers","title":"Make it ask before it answers","description":"Four lines that stop a model guessing at what you meant. It interviews you until it can actually do the task, which turns a mediocre first answer into a usable one.","category":"Fundamentals","tags":["prompting","accuracy","short","fundamentals"],"models":["claude","gpt","gemini","grok","perplexity"],"difficulty":"beginner","author":"Rony","version":"1.0","published":"2026-08-25","updated":"2026-08-25","variables":[{"name":"TASK","description":"What you want done, stated however roughly you like.","example":"Write the launch plan for our new pricing tier."}],"prompt":"Before answering, ask me clarifying questions until you are confident you can\ndo this correctly. Ask them one or two at a time, not as a list of ten.\n\nStop asking once more questions would not change your answer, then say so and\nproceed.\n\nDo not guess at anything I have not told you. Do not fill gaps with plausible\ndefaults. If you use an assumption anyway, label it inline as an assumption.\n\nTHE TASK:\n{{TASK}}","tips":["This is the single highest-leverage prompt in the library, and it's four lines. Most bad output is a guessing problem, not a model problem.","'One or two at a time' matters — a list of ten questions gets skimmed and half-answered.","If it asks nothing and proceeds, your task description was genuinely complete. That's worth knowing too."]},{"slug":"positioning-from-the-alternatives","url":"https://promptsbuddy.com/prompts/positioning-from-the-alternatives","title":"Position against what people actually do instead","description":"Most positioning compares you to competitors nobody evaluated. This works from the real alternative — a spreadsheet, an intern, or doing nothing — and finds the wedge from there.","category":"Marketing","tags":["positioning","marketing","strategy","messaging"],"models":["claude","gpt"],"difficulty":"advanced","author":"Rony","version":"1.0","published":"2026-08-25","updated":"2026-08-25","variables":[{"name":"PRODUCT","description":"What it is and what it does, plainly. No pitch.","example":"A reconciliation tool that flags mismatches between Stripe and your ledger."},{"name":"ALTERNATIVES","description":"What people genuinely do today — include 'nothing' and 'a spreadsheet' if true.","example":"A shared Google Sheet; a junior analyst two days a month; or living with the errors."},{"name":"WHO","description":"The specific person who feels the problem, not the buying committee.","example":"The one finance person at a 40-person company who owns month-end close."}],"prompt":"PRODUCT: {{PRODUCT}}\nWHAT THEY DO TODAY: {{ALTERNATIVES}}\nWHO FEELS IT: {{WHO}}\n\n**1. Rank the alternatives** by how many of these people actually use them.\nDoing nothing is an alternative and is usually the leader — say so if it is.\n\n**2. For each alternative**, give: what it genuinely does well, what it costs\nin time or risk, and the specific moment it breaks down. Be fair to it. If\na spreadsheet is fine for most of this job, say a spreadsheet is fine.\n\n**3. The wedge.** The one situation where the leading alternative fails badly\nenough that someone would change what they do. One sentence.\n\n**4. Three positioning statements** built on that wedge, each in the form\n\"For [who] who [situation], [product] is the [category] that [capability],\nunlike [alternative] which [limitation].\" Make the category name plain — if\nyou invent a category, say why an existing one won't do.\n\n**5. The strongest argument against us**, stated as the buyer would state it.\n\nRules:\n- Do not compare us to funded competitors unless I named them. The real\n  competitor is usually inertia.\n- No superlatives, no \"seamless\", no \"10x\".\n- Every claim must trace to what I told you. Write [NEEDS PROOF] rather than\n  inventing evidence.","tips":["If your positioning doesn't name 'doing nothing' as the main competitor, it's probably wrong.","Section 2's 'be fair to it' is what makes the output usable — conceding the true thing is what earns the rest.","[NEEDS PROOF] markers become your customer-research backlog."]},{"slug":"study-with-active-recall","url":"https://promptsbuddy.com/prompts/study-with-active-recall","title":"Study from your own notes with active recall","description":"Turns your messy notes into a testing session rather than a summary — it quizzes you, marks you honestly, and tells you which parts you only think you know.","category":"Learning","tags":["study","learning","revision","memory"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Meghna","version":"1.0","published":"2026-08-25","updated":"2026-08-25","variables":[{"name":"NOTES","description":"Your notes, however messy. Lecture scrawl is fine.","example":"Paste your notes."},{"name":"EXAM","description":"What you're being tested on and how, so the questions match.","example":"Written exam in 3 weeks, short-answer plus one essay."}],"prompt":"Here are my notes. Do not summarise them.\n\nNOTES:\n{{NOTES}}\n\nWHAT I'M PREPARING FOR: {{EXAM}}\n\nRun a recall session:\n\n1. Ask me **one question at a time** from the notes. Wait for my answer.\n2. Mark it honestly: correct / partially correct / wrong. For partial and\n   wrong, say exactly what was missing — do not just give the full answer.\n3. Ask a follow-up that tests whether I understood or only remembered wording.\n4. Vary difficulty based on how I'm doing. Return to anything I got wrong\n   later in the session, phrased differently.\n\nAfter ten questions, stop and give me:\n- **Solid** — what I clearly know.\n- **Fragile** — what I got right but explained shakily. This is the important list.\n- **Missing** — what I could not answer.\n- **Not in your notes** — anything the exam format implies I need that these\n  notes do not cover.\n\nRules:\n- Do not reveal the answer before I have attempted it.\n- Do not accept a vague answer as correct. Push once for specifics.\n- Do not be encouraging. Mark accurately.","tips":["Answer out loud or by typing before reading on. Recognising an answer isn't recalling it — that's the whole point.","The 'fragile' list is the study plan. Solid and missing both take care of themselves.","'Not in your notes' catches the gap between what you wrote down and what you'll be asked."]},{"slug":"clarity-rewrite-that-keeps-your-voice","url":"https://promptsbuddy.com/prompts/clarity-rewrite-that-keeps-your-voice","title":"Clarity rewrite that doesn't flatten your voice","description":"A deliberately short editing prompt: improves readability without changing structure, meaning, or the way you sound. The one people paste most often.","category":"Writing","tags":["editing","writing","clarity","short"],"models":["claude","gpt","gemini","grok"],"difficulty":"beginner","author":"Rony","version":"1.0","published":"2026-08-24","updated":"2026-08-24","variables":[{"name":"TEXT","description":"The paragraph or section to tighten.","example":"Paste the text."}],"prompt":"Rewrite the text below for clarity and flow.\n\nKeep my original voice, structure, and meaning. Improve readability only.\n\nDo not add new ideas. Do not change my argument. Do not remove hedges that\ncarry real uncertainty. Do not make it more formal than I wrote it.\n\nThen list, in one line each, the three changes you made that mattered most.\n\nTEXT:\n{{TEXT}}","tips":["Short on purpose. Longer editing prompts start rewriting your argument instead of your sentences.","The three-change list is how you learn to stop needing the prompt.","If the output sounds smoother but less like you, your text had voice in the hedges — put them back."]},{"slug":"cover-letter-that-isnt-generic","url":"https://promptsbuddy.com/prompts/cover-letter-that-isnt-generic","title":"A cover letter that couldn't be sent to anyone else","description":"Builds a short letter around one specific thing about the company and one specific thing you've done, then strips every sentence that would survive a find-and-replace of the company name.","category":"Career","tags":["cover-letter","job-search","career","writing"],"models":["claude","gpt"],"difficulty":"beginner","author":"Lena Fischer","version":"1.0","published":"2026-08-24","updated":"2026-08-24","variables":[{"name":"JOB_AD","description":"The posting.","example":"Paste the posting."},{"name":"MY_EVIDENCE","description":"Two or three things you have actually done that relate. Rough notes are fine.","example":"Rebuilt onboarding at a 12-person startup; cut time-to-first-value from 9 days to 2."},{"name":"WHY_THEM","description":"One genuine, specific reason this company — something you noticed, used, or read.","example":"I've used their API since 2024 and their changelog is the only one I actually read."}],"prompt":"Write a cover letter for this role. Under 200 words.\n\nPOSTING:\n{{JOB_AD}}\n\nWHAT I'VE ACTUALLY DONE:\n{{MY_EVIDENCE}}\n\nWHY THIS COMPANY:\n{{WHY_THEM}}\n\nStructure:\n- One opening sentence that is about them, not about me, and that could not\n  have been written by someone who hadn't looked.\n- One paragraph on the single most relevant thing I have done, with the\n  outcome, not the responsibility.\n- One sentence acknowledging the biggest gap between me and the posting, and\n  what I would do about it. Do not spin this.\n- One closing line proposing a concrete next step.\n\nThen run this check and show it: **would any sentence here survive if you\nswapped in a different company name?** List every sentence that would, because\nthose are the ones to cut or rewrite.\n\nRules:\n- Do not open with \"I am writing to apply for\". Do not use \"passionate\",\n  \"excited\", \"thrilled\", \"dynamic\", or \"fast-paced\".\n- Do not restate the résumé.\n- Do not flatter the company beyond the one specific thing I gave you.\n- If WHY THIS COMPANY is thin, say so rather than inventing enthusiasm.","tips":["The swap test is the whole prompt. Anything that survives it is filler.","Under 200 words is not a stylistic preference — long letters get skimmed to the first specific noun.","If you can't fill WHY_THEM honestly, that's information about whether to apply."]},{"slug":"interview-prep-from-the-job-ad","url":"https://promptsbuddy.com/prompts/interview-prep-from-the-job-ad","title":"Interview prep built from the job ad, not from a list","description":"Turns one posting into the eight questions you'll actually be asked, the two you're weakest on, and a drill for the story you keep telling badly.","category":"Career","tags":["interview","job-search","career","preparation"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Dinesh","version":"1.0","published":"2026-08-24","updated":"2026-08-24","variables":[{"name":"JOB_AD","description":"The posting.","example":"Paste the posting."},{"name":"MY_BACKGROUND","description":"Your résumé or a few paragraphs on what you've done.","example":"Paste your résumé."},{"name":"STAGE","description":"Which round this is — the questions differ a lot.","example":"Second round, with the hiring manager, 45 minutes."}],"prompt":"Prepare me for this interview. Round: {{STAGE}}.\n\nPOSTING:\n{{JOB_AD}}\n\nMY BACKGROUND:\n{{MY_BACKGROUND}}\n\nProduce:\n\n**1. Eight likely questions**, derived from the posting's specific\nresponsibilities and from the gaps between it and my background — not a generic\nlist. For each, name the thing the interviewer is actually testing.\n\n**2. My two weakest.** Where my background gives me the least to say. Be\ndirect about why.\n\n**3. One story, drilled.** Pick the single experience of mine most worth\ntelling for this role. Write it back to me in under 120 words: situation, the\ndecision I made, the outcome with a number if I gave you one, and what I would\ndo differently. Then list the two follow-up questions a sharp interviewer asks\nafter that story.\n\n**4. Five questions I should ask them**, that only make sense for this company\nand this role, and that would actually change whether I want the job.\n\nRules:\n- Do not invent details of my experience to make a story land. If the story is\n  thin, say the story is thin.\n- No generic advice about eye contact, STAR, or \"showing enthusiasm\".\n- Do not pad to eight questions if the posting only supports six. Say so.","tips":["Section 3 is where the value is. Most people have one story they tell badly and never rehearse.","The follow-up questions matter more than the story — that's where interviews are actually lost.","Ask for the questions before you write your answers, then answer cold. Reading answers first ruins the practice."]},{"slug":"match-my-writing-voice","url":"https://promptsbuddy.com/prompts/match-my-writing-voice","title":"Teach it your voice from three samples","description":"A short setup prompt: hand it three things you've written, have it derive the rules of how you write, then hold it to them for the rest of the session.","category":"Writing","tags":["voice","writing","style","short"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Meghna","version":"1.0","published":"2026-08-24","updated":"2026-08-24","variables":[{"name":"SAMPLE_ONE","description":"Something you wrote. 150+ words works better than a paragraph.","example":"Paste a real piece of your writing."},{"name":"SAMPLE_TWO","description":"A second sample, ideally in a different register.","example":"A Slack message or an email you sent."},{"name":"SAMPLE_THREE","description":"A third. More range means better rules.","example":"A doc or a post."}],"prompt":"Here are three things I wrote.\n\nSAMPLE 1:\n{{SAMPLE_ONE}}\n\nSAMPLE 2:\n{{SAMPLE_TWO}}\n\nSAMPLE 3:\n{{SAMPLE_THREE}}\n\nFirst, derive the rules of how I write. Be specific and observable — average\nsentence length, how I open, whether I use questions, which punctuation I\nfavour, words I reach for, words I clearly avoid, how I handle uncertainty,\nhow I end things. Ten rules maximum. Quote me as evidence for each.\n\nThen state the two things about my writing that are weaknesses, not style.\n\nFrom that point on, write everything in this conversation to those rules.\n\nDo not flatter the samples. Do not describe my voice with adjectives like\n\"engaging\" or \"conversational\" — give me rules I could check a draft against.","tips":["Samples beat adjectives. 'Professional but friendly' means nothing; 450 words of your writing means everything.","Pick samples in different registers — an email and an essay — or it learns the format, not the voice.","Save the ten rules. They're reusable, and they're a decent style guide for anyone writing on your behalf."]},{"slug":"tailor-a-resume-to-one-job","url":"https://promptsbuddy.com/prompts/tailor-a-resume-to-one-job","title":"Tailor a résumé to one specific job, honestly","description":"Maps your real experience against a single job ad, rewrites the bullets that match, and tells you plainly which requirements you don't meet instead of papering over them.","category":"Career","tags":["resume","job-search","career","writing"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Rony","version":"1.0","published":"2026-08-24","updated":"2026-08-24","variables":[{"name":"RESUME","description":"Your current résumé, pasted as plain text.","example":"Paste the whole thing, formatting and all."},{"name":"JOB_AD","description":"The full job posting, including the boring requirements section.","example":"Paste the posting verbatim."}],"prompt":"Here is my résumé and one job posting. Tailor the résumé to the posting\nwithout inventing anything.\n\nRESUME:\n{{RESUME}}\n\nJOB POSTING:\n{{JOB_AD}}\n\nWork in this order:\n\n**1. The match table.** Every requirement in the posting, in its own row, with:\nthe evidence in my résumé that satisfies it (quote it), and a rating of\nstrong / partial / absent. Be strict. \"Adjacent experience\" is partial, not strong.\n\n**2. Gaps.** The requirements rated absent. For each, say whether it is likely\na hard filter or a nice-to-have, based on how the posting words it.\n\n**3. The rewrite.** Rewrite only the bullets where the match is strong or\npartial, so they lead with the outcome and use the posting's own vocabulary\nwhere my experience genuinely fits it. Keep every number I gave you unchanged.\n\n**4. What to cut.** Bullets that earn no space against this specific posting.\n\nRules:\n- Do not invent experience, tools, numbers, or job titles. If a bullet has no\n  metric, leave it without one rather than estimating.\n- Do not restate a requirement as if it were experience.\n- Keep my voice. If I write plainly, do not make it corporate.\n- Where a gap is real, say so in one sentence and move on. No reassurance.","tips":["Run it per job, not per company. The whole value is in the posting-specific vocabulary.","The match table is the useful output even if you don't use the rewrite — it tells you whether to apply at all.","If everything comes back 'strong', your résumé is too vague to evaluate. That's the finding."]},{"slug":"turn-a-transcript-into-a-decision-doc","url":"https://promptsbuddy.com/prompts/turn-a-transcript-into-a-decision-doc","title":"Turn a messy meeting transcript into a decision doc","description":"Separates what was decided from what was merely discussed, attributes owners, and flags the open questions everyone left the room assuming someone else would answer.","category":"Product","tags":["meetings","documentation","product","async"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Lena Fischer","version":"1.2","published":"2026-07-21","updated":"2026-07-21","variables":[{"name":"TRANSCRIPT","description":"Raw transcript, notes, or a chat log. Speaker labels help but aren't required.","example":"Paste the transcript."},{"name":"PURPOSE","description":"What the meeting was for.","example":"Deciding whether to rebuild search or buy a vendor."},{"name":"READER","description":"Who reads the doc — this changes what's worth including.","example":"Two engineers who weren't there, plus our VP who will skim it."}],"prompt":"Turn this raw meeting record into a decision document for {{READER}}.\n\nMEETING PURPOSE: {{PURPOSE}}\n\nTRANSCRIPT:\n{{TRANSCRIPT}}\n\nProduce exactly these sections:\n\n**Decisions made.** Only things actually settled. For each: the decision in one\nsentence, who owns it, and the date or trigger it takes effect. If a decision was\nimplied but never stated out loud, put it here and mark it `[implied — confirm]`.\n\n**Explicitly not decided.** Things discussed at length that reached no\nconclusion. This section is the most valuable one; do not let it be empty if the\ntranscript shows unresolved threads.\n\n**Open questions.** Each phrased as a question, with a named owner and the\ndecision it blocks. If no owner was named, write `owner: unassigned` — do not guess.\n\n**Context worth keeping.** The two or three pieces of reasoning a future reader\nwould need to understand why, not what. Discard the rest.\n\n**Disagreements.** Where people did not converge, state both positions fairly and\nwho held each. Do not smooth this over into consensus.\n\nRules:\n- Never invent an owner, a date, or a decision. Absence is information.\n- Quote directly when the exact wording carries the nuance.\n- Cut small talk, tangents, and anything already in a shared doc.\n- Total length: under one screen for the first three sections.","tips":["'Explicitly not decided' is the section that stops the same meeting happening twice.","Send the draft into the meeting channel within an hour. Corrections arrive while memory is fresh.","For recurring meetings, keep the output format identical — the diff between weeks becomes the real status report."]},{"slug":"sql-from-a-vague-question","url":"https://promptsbuddy.com/prompts/sql-from-a-vague-question","title":"Vague executive question → SQL that survives review","description":"Forces the ambiguity out of a business question before any SQL gets written, then produces a query with its assumptions stated on the record.","category":"Data","tags":["sql","analytics","data","requirements"],"models":["claude","gpt"],"difficulty":"advanced","author":"Sam Reyes","version":"2.2","published":"2026-07-09","updated":"2026-08-19","variables":[{"name":"QUESTION","description":"The question as it was asked, in the asker's words.","example":"How many of our customers are actually active?"},{"name":"SCHEMA","description":"DDL or a column listing for the relevant tables.","example":"users(id, created_at, deleted_at), events(user_id, name, ts), accounts(id, plan, churned_at)"},{"name":"DIALECT","description":"Warehouse and dialect.","example":"BigQuery, standard SQL"}],"prompt":"A stakeholder asked: \"{{QUESTION}}\"\n\nSCHEMA:\n{{SCHEMA}}\n\nDIALECT: {{DIALECT}}\n\nWork in three steps and show all three.\n\n**Step 1 — Ambiguities.** List every word in the question that could be defined\nmore than one way, with the plausible definitions and which one you'll use.\n(\"Active\" alone usually has four.) For each, note whether the choice would\nmaterially change the number.\n\n**Step 2 — The definition.** Restate the question as a precise, testable one,\nin the form: \"Count of X where Y, over period Z, excluding W.\" This is what you\nwill actually answer.\n\n**Step 3 — The query.** Write it for {{DIALECT}}. Requirements:\n- Every non-obvious choice gets an inline comment saying why.\n- CTEs named for what they mean, not `t1`.\n- Handle nulls, soft deletes, and timezone explicitly — state the timezone.\n- No `SELECT *`.\n- If the schema can't answer the question, say exactly what's missing and write\n  the closest honest query instead. Do not invent columns.\n\nThen add:\n- **Sanity checks** — two queries whose results should hold if the main query is\n  right (row-count bounds, a known subtotal, a spot-check on one user).\n- **How this gets misread** — the one way a reader will misinterpret the number,\n  and the caption to prevent it.","tips":["Step 1 is the whole prompt. Most 'wrong' dashboards are correct SQL answering a question nobody agreed on.","Send Step 2 back to the stakeholder before you run anything. Two lines of confirmation saves a day of rework.","Keep the sanity checks in the PR — reviewers approve queries far faster when they can see them pass."]},{"slug":"pre-mortem-for-any-plan","url":"https://promptsbuddy.com/prompts/pre-mortem-for-any-plan","title":"The pre-mortem: find how your plan fails before it does","description":"Hands the model a plan and asks it to write the post-mortem for a failure that hasn't happened yet. Surfaces the risks your team is too invested to name out loud.","category":"Strategy","tags":["planning","risk","product","decision-making"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Rony","version":"1.3","published":"2026-07-02","updated":"2026-08-14","variables":[{"name":"PLAN","description":"Paste the plan, PRD, or launch doc. Rough notes are fine.","example":"Ship self-serve onboarding by Q4 so sales stops doing demos for sub-$5k accounts."},{"name":"HORIZON","description":"How far out the failure is imagined.","example":"six months"},{"name":"CONTEXT","description":"Team size, constraints, anything the model can't infer.","example":"Four engineers, one designer, no dedicated PM. Enterprise contracts renew in Q1."}],"prompt":"You are running a pre-mortem. It is {{HORIZON}} from now and the plan below has\nfailed — not spectacularly, but in the quiet, ordinary way real plans fail.\n\nPLAN:\n{{PLAN}}\n\nCONTEXT:\n{{CONTEXT}}\n\nWrite the post-mortem for that failure. Follow this structure exactly:\n\n1. **The headline.** One sentence a colleague would say in the hallway.\n2. **What actually happened.** A short narrative, in past tense, of how it went\n   wrong. Be specific about sequence and timing.\n3. **The five root causes.** For each: name it, say which assumption in the plan\n   it violates, and rate how visible it was at the start (obvious / knowable /\n   genuinely unforeseeable).\n4. **The tell.** For each root cause, the earliest observable signal that would\n   have caught it — something measurable, not a feeling.\n5. **What we'd change on day one.** Three concrete edits to the plan as written.\n\nRules:\n- Do not hedge. Commit to a specific failure story rather than listing generic risks.\n- At least two root causes must be organisational or human, not technical.\n- Name the thing everyone would have been too polite to say in the kickoff meeting.\n- If the plan is missing information you'd need, state the assumption you made\n  inline rather than asking me.","tips":["Run it twice with different HORIZON values. Six-month failures are execution problems; two-year failures are strategy problems.","The 'tell' section is the real output — those become your dashboard metrics.","Paste the model's five root causes back into your planning doc as open questions, not as answers."]},{"slug":"design-critique-partner","url":"https://promptsbuddy.com/prompts/design-critique-partner","title":"A design critique partner that argues back","description":"Critiques a screen or flow against stated goals, then argues the opposite position — so you get a real second opinion instead of polite agreement.","category":"Design","tags":["design","critique","ux","feedback"],"models":["claude","gpt","gemini"],"difficulty":"intermediate","author":"Lena Fischer","version":"1.6","published":"2026-06-25","updated":"2026-06-25","variables":[{"name":"ARTEFACT","description":"Describe the screen, paste the copy, or attach the image.","example":"Checkout step 2: shipping form, 9 fields, single-column, progress bar at top."},{"name":"GOAL","description":"What this screen must achieve, and for whom.","example":"Get a first-time buyer through shipping in under 40 seconds on mobile."},{"name":"CONSTRAINTS","description":"What can't change — legal, technical, brand.","example":"Address validation is required by our carrier. Cannot remove the phone field."}],"prompt":"Act as a senior designer giving critique in a review, not a consultant writing a\nreport.\n\nTHE WORK:\n{{ARTEFACT}}\n\nGOAL: {{GOAL}}\nFIXED CONSTRAINTS: {{CONSTRAINTS}}\n\n**Round 1 — Critique.** Five observations, ordered by how much they affect the\ngoal. For each:\n- What you see (describe it neutrally, no judgement yet)\n- The user consequence, stated as a behaviour, not a feeling\n- How confident you are, and what would settle it (a test, a metric, a session)\n\nDo not suggest fixes yet.\n\n**Round 2 — The counter-argument.** Take your single strongest criticism and\nargue convincingly that it is wrong: the designer had a reason you didn't see.\nBe genuinely persuasive. Then say which side you actually land on and why.\n\n**Round 3 — Two directions.** Not a list of tweaks. Two coherent alternative\napproaches, each with what it optimises for, what it sacrifices, and who would\nhate it. Respect the fixed constraints absolutely.\n\nRules:\n- No praise sandwich. If something works, say so once, specifically.\n- Never cite a \"best practice\" without saying what it's trading off.\n- Anything you can't judge from what I gave you: say so plainly rather than\n  guessing at the visual design.","tips":["Round 2 is what makes this useful. Without it the model just agrees with whatever framing you brought.","Feed it the goal in measurable terms. 'Make it cleaner' produces critique about whitespace; '40 seconds on mobile' produces critique about field count.","Works on your own week-old work — you'll defend the counter-argument out loud and learn what you actually believe."]},{"slug":"code-review-that-teaches","url":"https://promptsbuddy.com/prompts/code-review-that-teaches","title":"A code review that teaches instead of nitpicks","description":"Reviews a diff at three altitudes — design, correctness, style — and explains the reasoning behind each note so the author learns something instead of just applying a patch.","category":"Engineering","tags":["code-review","engineering","mentorship","quality"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Meghna","version":"2.1","published":"2026-06-18","updated":"2026-08-09","variables":[{"name":"DIFF","description":"The diff or full file. Include surrounding context if the change is small.","example":"git diff main...HEAD"},{"name":"LANGUAGE","description":"Language and framework, so idiom advice is right.","example":"TypeScript, Next.js App Router"},{"name":"STANDARDS","description":"Anything the model can't infer: house style, perf budgets, review norms.","example":"We prefer explicit over clever. No new dependencies without discussion."}],"prompt":"Review the following {{LANGUAGE}} change the way a senior engineer reviews a\ncolleague's work: honestly, specifically, and with the reasoning shown.\n\nHOUSE STANDARDS:\n{{STANDARDS}}\n\nDIFF:\n{{DIFF}}\n\nStructure the review in three passes, in this order:\n\n**Pass 1 — Design.** Is this the right shape of solution? Does it belong here?\nWhat will this make harder in six months? Skip if the change is trivial, and say\nso rather than inventing concerns.\n\n**Pass 2 — Correctness.** Bugs, race conditions, unhandled errors, boundary\ncases, security. For each: give the concrete input or state that triggers it and\nthe resulting wrong behaviour. If you cannot name the trigger, you are guessing —\ndrop the note.\n\n**Pass 3 — Craft.** Naming, structure, idiom, tests. Only things that would\nsurvive a \"does this matter?\" challenge.\n\nFor every note, use this format:\n  - **What:** the observation, one line.\n  - **Why:** the principle behind it, one line. Not \"best practice\" — the actual reason.\n  - **Fix:** a code suggestion, or \"author's call\" if it's genuinely taste.\n\nFinally, add a section called **What's good here** with two specific things done\nwell, and a one-line verdict: approve / approve with comments / needs work.\n\nDo not restate what the code does. Do not comment on formatting a linter handles.","tips":["Feed it the whole file when the diff is under ~40 lines — reviews of isolated hunks invent problems that the surrounding code already solves.","The 'Why' line is the teaching. If it reads like a slogan, push back: 'explain why, from first principles.'","Keep STANDARDS in a snippet file and paste it every time. Consistency across reviews is most of the value."]},{"slug":"support-reply-that-de-escalates","url":"https://promptsbuddy.com/prompts/support-reply-that-de-escalates","title":"The support reply that de-escalates without grovelling","description":"Drafts a response to an angry customer that acknowledges the real complaint, commits to something specific, and never apologises for things that aren't your fault.","category":"Support","tags":["support","customer","writing","communication"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Dinesh","version":"2.0","published":"2026-06-04","updated":"2026-08-11","variables":[{"name":"MESSAGE","description":"The customer's message, verbatim.","example":"Third time this month your export has failed. We're evaluating alternatives."},{"name":"FACTS","description":"What actually happened, including anything you can't say publicly.","example":"Known bug in the CSV writer for >50k rows. Fix ships Thursday. Not disclosed publicly yet."},{"name":"AUTHORITY","description":"What you're actually allowed to offer.","example":"Can offer a credit up to one month. Cannot promise dates for unshipped work."}],"prompt":"Draft a reply to this customer.\n\nCUSTOMER MESSAGE:\n{{MESSAGE}}\n\nWHAT I ACTUALLY KNOW:\n{{FACTS}}\n\nWHAT I CAN OFFER:\n{{AUTHORITY}}\n\nBefore writing, answer these three internally and show them as a short block\nlabelled ANALYSIS:\n1. What is the customer's stated complaint?\n2. What is the underlying one? (Usually: lost time, lost credibility with their\n   own boss, or fear this will keep happening.)\n3. What outcome would make them consider this resolved?\n\nThen write the reply. Requirements:\n- Open by naming the actual impact on them, not by apologising. One sentence.\n- State what happened in plain language. No passive voice hiding the actor.\n- Commit to exactly one specific next step with a date or a named person.\n- Never promise anything outside WHAT I CAN OFFER. If they asked for something\n  you can't give, say no clearly and say what you can do instead.\n- No \"we sincerely apologise for any inconvenience\". No \"we value your feedback\".\n- Match their register: if they wrote three terse lines, don't send eight paragraphs.\n- End without a question unless you genuinely need an answer to proceed.\n\nAfter the reply, add a line: **Risk** — the one thing in this message that could\nmake it worse, and what you'd change if the customer is more senior than assumed.","tips":["The ANALYSIS block is for you, not the customer. Read it, then delete it before sending.","'Never apologise first' surprises people. Naming the impact reads as understanding; a lead apology reads as a script.","If the model keeps sneaking in a promise you can't keep, tighten AUTHORITY — that field is doing the guardrail work."]},{"slug":"rewrite-until-it-sounds-human","url":"https://promptsbuddy.com/prompts/rewrite-until-it-sounds-human","title":"Rewrite until it stops sounding like AI wrote it","description":"A diagnostic-then-edit pass that names the specific tells in a draft — the tricolons, the hedges, the empty transitions — and fixes them without flattening your voice.","category":"Writing","tags":["writing","editing","voice","content"],"models":["claude","gpt"],"difficulty":"beginner","author":"Rony","version":"1.8","published":"2026-05-30","updated":"2026-08-20","variables":[{"name":"DRAFT","description":"The text to fix. Works best under 800 words.","example":"Paste your draft here."},{"name":"VOICE","description":"Two or three writers, publications, or samples that sound like you want to sound.","example":"Plain and direct, like a good internal memo. No jokes, no metaphors."},{"name":"AUDIENCE","description":"Who reads it and what they already know.","example":"Engineering managers who have read a hundred of these."}],"prompt":"Two tasks, in order. Do not skip the first.\n\n**Task 1 — Diagnose.** Read the draft below and list every specific place it\nreads as machine-written. For each, quote the exact phrase and name the pattern.\nLook for:\n- Three-item lists where two items would do\n- Hedges that remove all commitment (\"can help you\", \"may be able to\", \"often\")\n- Transitions that carry no information (\"moreover\", \"in today's fast-paced\")\n- Abstract nouns doing work a verb should do (\"the implementation of\")\n- Sentences that restate the previous sentence in different words\n- Grand closing paragraphs that summarise rather than land\n- Uniform sentence length — real writing varies hard\n\nIf a passage is already good, say so and leave it alone. Do not manufacture\nproblems.\n\n**Task 2 — Rewrite.** Produce a clean version that fixes everything in Task 1\nwhile keeping the author's actual voice.\n\nVOICE TARGET: {{VOICE}}\nAUDIENCE: {{AUDIENCE}}\n\nRules for the rewrite:\n- Keep every fact, number, and claim. Change none of them.\n- Cut length by 15–30%. If you can't, say why.\n- Vary sentence length deliberately. Short sentences carry the important parts.\n- Prefer the concrete noun to the category (\"Postgres\", not \"the database layer\").\n- Keep any sentence where the author's personality shows, even if it's slightly awkward.\n- No em-dash pileups, no rhetorical questions used as transitions.\n\nEnd with a two-line note on what you changed most and one thing the author\nshould decide themselves.\n\nDRAFT:\n{{DRAFT}}","tips":["The diagnosis is the reusable part. After ten drafts you stop making those mistakes and stop needing the prompt.","If the rewrite loses your voice, the VOICE field was too abstract. Paste 200 words of your own writing instead of adjectives.","Never accept the rewrite wholesale. Take the cuts, keep your own sentences where they were already fine."]},{"slug":"landing-page-copy-from-customer-words","url":"https://promptsbuddy.com/prompts/landing-page-copy-from-customer-words","title":"Landing page copy built from your customers' own words","description":"Mines real customer language from reviews, tickets, and calls, then writes a page that uses their phrasing instead of your internal vocabulary.","category":"Marketing","tags":["copywriting","marketing","positioning","landing-page"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Rony","version":"1.4","published":"2026-05-12","updated":"2026-05-12","variables":[{"name":"RAW_VOICE","description":"Paste 10–30 real quotes: reviews, support tickets, sales call notes, Reddit threads.","example":"\"I spent my whole Sunday reconciling two spreadsheets that should have matched.\""},{"name":"PRODUCT","description":"What it is and what it does, plainly.","example":"A reconciliation tool that flags mismatches between Stripe and your ledger."},{"name":"ALTERNATIVE","description":"What they do today instead — including 'nothing' and 'a spreadsheet'.","example":"A shared Google Sheet and a monthly panic."}],"prompt":"You are writing a landing page using only language the customers actually use.\n\nRAW CUSTOMER VOICE:\n{{RAW_VOICE}}\n\nPRODUCT: {{PRODUCT}}\nWHAT THEY DO TODAY: {{ALTERNATIVE}}\n\n**Step 1 — Extract.** From the raw voice, pull out:\n- The exact phrases people use for the problem (verbatim, in quotes)\n- The words they use for the outcome they want\n- Any word *we* use that they never use — flag these as vocabulary to drop\n- The emotional register: is this annoying, embarrassing, or frightening to them?\n\n**Step 2 — Write the page.** Sections: hero (headline + subhead + one CTA), the\nproblem in their words, how it works in three steps, one objection handled\nhead-on, and a closing CTA.\n\nConstraints:\n- The headline must contain at least one phrase lifted from the raw voice.\n- No superlatives. No \"revolutionary\", \"seamless\", \"powerful\", \"unlock\".\n- Name the alternative explicitly and say honestly where it's fine. Trust is\n  built by conceding the true thing.\n- Reading level: a busy person skimming on a phone. Sentences under 20 words.\n- Every claim must be traceable to the raw voice or the product description. If\n  you need a claim you can't support, write `[NEEDS PROOF]` instead of inventing it.\n\n**Step 3 — Three headline alternatives**, each taking a different angle:\nthe pain, the outcome, and the enemy.","tips":["Garbage in, garbage out: 5 vague quotes produce a generic page. 25 real ones produce copy you couldn't have written.","The 'vocabulary to drop' list is worth pinning up. It's your internal jargon, visible for the first time.","[NEEDS PROOF] markers become the customer-story backlog for your marketing team."]},{"slug":"learn-one-hard-idea","url":"https://promptsbuddy.com/prompts/learn-one-hard-idea","title":"Learn one hard idea properly in an hour","description":"Builds a one-hour path through a concept you keep bouncing off: the minimum prerequisite, the idea itself, the two places your intuition will be wrong, and a test.","category":"Learning","tags":["learning","teaching","study","explanation"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Meghna","version":"1.2","published":"2026-05-06","updated":"2026-05-06","variables":[{"name":"IDEA","description":"The concept you keep failing to hold onto.","example":"Why database isolation levels matter in practice."},{"name":"BACKGROUND","description":"What you already know, honestly. Being specific here changes everything.","example":"I write application code against Postgres daily. I know what a transaction is. I have never tuned one."},{"name":"PURPOSE","description":"Why you need it — this decides the depth.","example":"So I can review a colleague's change to our payment writes without nodding along."}],"prompt":"Teach me this in about an hour: {{IDEA}}\n\nWHERE I'M STARTING FROM: {{BACKGROUND}}\nWHY I NEED IT: {{PURPOSE}}\n\nBuild the session in this shape:\n\n**1. The one prerequisite.** The single thing I must understand first, explained\nin a paragraph. If I already have it from my background, say so and skip.\n\n**2. The idea itself.** Explain it once concretely, using an example from my own\ndomain based on what I told you. Then explain it once more abstractly. Concrete\nfirst, always.\n\n**3. Where my intuition will be wrong.** Two specific misconceptions someone with\nmy background reliably has, why they're tempting, and what actually happens.\n\n**4. The boundary.** When this idea stops applying, and what replaces it.\n\n**5. Ten minutes of practice.** One exercise I can do right now with what's on my\nmachine. Give the task, not the answer.\n\n**6. Check.** Three questions. If I can answer them, I have it. Include what a\nwrong answer to each reveals.\n\nRules:\n- No analogies to trains, restaurants, or libraries unless the mechanism genuinely matches.\n- Do not flatter the difficulty (\"this trips up a lot of people!\"). Just teach.\n- If the idea is genuinely too large for an hour, say so and teach the most\n  load-bearing third instead of skimming all of it.","tips":["Be honest in BACKGROUND. Overstating it gets you an explanation you'll nod through and forget.","Do the exercise before reading section 6. Reading the check questions first turns them into hints.","Save section 3 somewhere. Misconception lists are the part worth rereading in six months."]},{"slug":"explain-this-codebase-to-me","url":"https://promptsbuddy.com/prompts/explain-this-codebase-to-me","title":"Explain this codebase like I start on Monday","description":"Produces the orientation doc a new engineer actually needs: the three things that will confuse them, the load-bearing files, and the conventions nobody wrote down.","category":"Engineering","tags":["onboarding","engineering","documentation","codebase"],"models":["claude","gpt"],"difficulty":"intermediate","author":"Meghna","version":"1.5","published":"2026-04-27","updated":"2026-04-27","variables":[{"name":"TREE","description":"Output of `tree -L 3 -I node_modules` or similar.","example":"app/\n  api/\n  (marketing)/\nlib/\ncomponents/"},{"name":"ENTRY","description":"Two or three key files, pasted in full.","example":"app/layout.tsx, lib/db.ts, package.json"},{"name":"ROLE","description":"What the reader will be doing first.","example":"Adding a new API route and its tests."}],"prompt":"I am joining this codebase on Monday and my first task is: {{ROLE}}.\nWrite the orientation I wish someone would give me over coffee.\n\nSTRUCTURE:\n{{TREE}}\n\nKEY FILES:\n{{ENTRY}}\n\nCover these, in order:\n\n1. **What this system does**, in three sentences, in terms of what a user\n   experiences — not architecture.\n2. **The shape of it.** The 4–6 pieces that matter and how a request flows\n   through them. One paragraph, no diagrams-in-text.\n3. **Where I'll actually work.** Given my first task, the specific files I'll\n   touch and the order I should read them.\n4. **Three things that will confuse me.** Non-obvious conventions, naming that\n   means something different here, places where the code disagrees with itself.\n   Be specific and cite files.\n5. **The load-bearing files.** What breaks everything if edited carelessly.\n6. **What I should not touch yet**, and why.\n7. **Five questions to ask the team**, ones the code cannot answer — history,\n   intent, or plans.\n\nRules:\n- Mark inference explicitly: write `[inferred]` when reasoning from structure\n  rather than from code you were shown.\n- Do not summarise files one by one. Synthesise.\n- Prefer \"this is unusual\" over \"this follows best practice\" — the unusual parts\n  are what I need warned about.","tips":["Section 7 is the highest-value output. Take those five questions to your team lead in week one.","Run it again after two weeks with your own corrections pasted in. The delta shows you what you actually learned.","Add `git log --format='%s' -n 50` to the context and the model picks up on where the churn is."]},{"slug":"weekly-review-that-actually-sticks","url":"https://promptsbuddy.com/prompts/weekly-review-that-actually-sticks","title":"The weekly review that survives a bad week","description":"Turns a week of scattered notes into an honest account of where the time went, what moved, and the one thing worth protecting next week.","category":"Career","tags":["productivity","reflection","career","planning"],"models":["claude","gpt","gemini"],"difficulty":"beginner","author":"Rony","version":"1.1","published":"2026-04-15","updated":"2026-04-15","variables":[{"name":"RAW","description":"Calendar exports, todo dumps, notes, commit messages, Slack you starred. Messy is fine.","example":"Mon: 4 meetings. Shipped the export fix. Tue: interviews all day..."},{"name":"GOALS","description":"The two or three things that were supposed to matter this quarter.","example":"Ship self-serve onboarding. Hire one senior engineer. Stop being the bottleneck on reviews."}],"prompt":"Here is a raw record of my week. Help me see it honestly.\n\nRAW RECORD:\n{{RAW}}\n\nWHAT WAS SUPPOSED TO MATTER THIS QUARTER:\n{{GOALS}}\n\nWrite:\n\n**Where the time went.** Rough percentages across categories you infer from the\nrecord. Name the largest category plainly, even if it's meetings or firefighting.\n\n**Against the goals.** For each stated goal: moved / stalled / not touched, and\nthe specific evidence from the record. Do not be generous with \"moved\".\n\n**The gap.** One paragraph on the distance between what I said mattered and\nwhere the hours actually went. No pep talk. If they matched, say so — that's\nworth knowing too.\n\n**Three things that worked.** Specific, from the record, worth repeating.\n\n**One thing to protect next week.** A single commitment, with the specific hours\nblocked and the specific thing you'd say no to in order to keep it. Just one.\n\nRules:\n- Use only what's in the record. If it isn't there, it didn't happen — say the\n  record is thin rather than filling gaps.\n- No motivational language. No \"great job on\". Flat and factual.\n- If the week was genuinely bad, say that in one sentence and move on.","tips":["Feed it your calendar export, not your memory. Memory over-weights the interesting hours.","Keep the outputs in one file. Twelve weeks of 'the gap' paragraphs is the most useful career document you'll own.","'One thing to protect' works because it's one. Three commitments is a wish list."]}]}