Skip to content
llmwise
Sign inStart free

Guide · Prompts

AI prompt library

30 prompts for everyday work, from rewriting an email to reviewing code. Copy one, replace the parts in [brackets], and send it to any model in llmwise.

Start freeSee pricing

How to use them

  1. Replace every part in [brackets] with your own details. The more specific you are, the better the answer.

  2. Paste the material itself, or attach the file, rather than describing it.

  3. Start on an everyday model (Claude Haiku 5.5, GPT-6 Luna, DeepSeek V4.1 Flash, and GLM 5.3 Flash). Move up to a bigger model when an answer isn't good enough.

Writing

  • Rewrite for clarity

    Rewrite the text below so it's clear and plain. Keep my meaning and my facts, cut filler, and keep it under [number] words. Show only the rewrite. [paste your text]

    Which models, and what a run costs

  • Draft an email

    Write an email to [who] about [what]. What I need from them: [the ask]. Tone: [friendly / formal]. Under 150 words, with the ask in the last line.

  • Feedback, not a rewrite

    Read my draft below as [the audience]. List the three changes that would improve it most, most important first, quoting the sentence each one applies to. Don't rewrite it. [paste your draft]

  • Headline options

    Suggest ten headlines for the piece below, for [where it will appear]. Mix plain and punchy, no clickbait. Mark your top three and say why. [paste the piece]

Summaries and documents

Attach the file to the message: PDFs, images and text files.

  • A summary for a busy reader

    Summarize the attached document for [who]. Start with the one-sentence takeaway, then up to five bullet points with the numbers that matter, then anything I need to decide or do.

    Which models, and what a run costs

  • Pull out the facts

    From the attached file, make a table of [the items: dates, amounts, names], with the page each one comes from. If something is unclear, say so instead of guessing.

    Which models, and what a run costs

  • Compare two versions

    Compare the two attached versions. List what changed, what was added and what was removed, and flag anything that changes the meaning.

  • Meeting notes to actions

    Turn these notes into three lists: decisions made, action items (with owner and date), and open questions. [paste your notes]

    Which models, and what a run costs

Coding

  • Explain this code

    Explain what this code does, step by step, for someone who knows the basics of [language]. Then list anything that looks like a bug or a risk. [paste the code]

  • Fix an error

    Here's my code and the error it gives. Explain the cause in one or two sentences, then show the smallest change that fixes it. [paste the code] [paste the error]

  • Write tests

    Write unit tests for this function with [test framework]. Cover the normal case, the edge cases and bad input: one test per behaviour, named after it. [paste the function]

  • Review a change

    Review this change like a careful senior engineer: correctness first, then readability. Quote the line each point is about, and skip style nits. [paste the diff]

Data and spreadsheets

  • Make sense of a CSV

    The attached CSV is [what it is]. Tell me the five most important things in it, with the numbers behind each, and the one chart you'd make to show them.

  • Write a formula

    In a spreadsheet I have [the columns]. Write a formula that [does what], explain each part, and show it working on a few made-up rows.

  • Build a comparison table

    Make a spreadsheet comparing [the options] on [the criteria], one row each, with a short note in each cell and a last column with your pick and why.

  • Check my reasoning

    Here's my analysis and my conclusion. Point out any step that doesn't follow, any number that looks off, and what data would change the conclusion. [paste your analysis]

Learning

  • Explain it three ways

    Explain [topic] three ways: in one sentence, in a paragraph for a curious 15-year-old, and in depth for someone who knows [a related field].

  • Quiz me

    Quiz me on [topic], one question at a time. Wait for my answer, tell me whether I'm right and why, then ask a slightly harder question.

  • A study plan

    I have [number] weeks to learn [skill], with [hours] a week. Make a week-by-week plan with one practical exercise each week.

  • Both sides, at their strongest

    Make the strongest case for and against [position], each the way someone who believes it would put it. Then name the one fact each side most needs to be true.

Research

These ask the model to search the web; it shows its sources as cards. On Free, web search works inside your 5 trial messages.

  • An answer with sources

    Search the web for [question]. Answer in five bullet points, each with its source, and say where the sources disagree.

    Which models, and what a run costs

  • Read a page for me

    Read [link] and tell me the three main claims it makes, the evidence for each, and anything important it leaves out.

  • Compare products

    Search for the current [kind of product] options under [budget]. Make a table of the top five with price, key specs and one downside each, with sources.

Making things

These make something in the side panel: a document or a deck you can edit and download, or a small app that runs beside the chat.

  • A small tool

    Make a [calculator / tracker / quiz] as a small HTML app that [does what]. Keep it simple, and let me change [the inputs].

  • A slide deck

    Make a [number]-slide deck on [topic] for [audience]: a title slide, one idea per slide, and a closing slide with next steps.

  • A one-page document

    Write a one-page [plan / brief / proposal] for [goal] as a document, with headings, and end with the three open questions I should settle first.

Tested prompt packs

Prompts for one topic each, every one run on three or four models, with the replies side by side, which passed our check, and what each cost.

Every model's reply to one prompt

One prompt to a page, sent to every model in llmwise: each reply as it came, whether it passed its check, and what it cost. 50 from our own test runs, and 50 test inputs of the free tools.

From our test runs

From the free tools' test runs

Prompts, priced model by model

Prompts that need something particular from a model (a file it must read, a tool, a long paste), each with the models that can run it and what one run counts as on each.

Make any prompt better

Say who it's for, give an example of what good looks like, and ask for the format you'll use. More in prompt optimization.

Questions

Do these prompts work with every model?

Yes. They're plain instructions, not tuned to one model. Start on an everyday model; if the answer isn't good enough, switch models in the same chat and ask again.

How do I reuse a prompt I like?

Save it as a persona in Settings → Skills: a name, the instructions, and a default model. Then pick that persona in the message box instead of pasting the prompt again.

Does a longer prompt cost more?

Not by itself: each message has a fixed price, shown before you send. Only once a whole chat sent to the model passes 64k tokens does a message count as 2 (past 128k, as 4), and the message box says so first.

Claude, GPT, Gemini, DeepSeek, Grok, Kimi, GLM, and Mistral, in one chat.

See what a message costs before you send it. Free is 5 messages to try; sign in with an email link, no password or card.

Start freeSee pricing