Why keep a prompt library at all
Most people who use AI assistants regularly end up writing roughly the same handful of prompts over and over, with small variations — a follow-up email template, a "summarize this into action items" request, a "quiz me on this topic" study prompt. Rewriting each one from scratch every time is a small but real tax on how useful the tool feels day to day. A prompt library exists to eliminate that tax: a categorized, searchable set of prompts that already reflect good structure, ready to copy and adapt rather than draft cold.
The anatomy of a reusable prompt
Every prompt in this library follows the same underlying structure, which mirrors the checklist covered in more depth on our Prompt Length Estimator page: a clearly stated task, an implicit or explicit format expectation, and — where relevant — a placeholder for the specific context that makes the prompt yours rather than generic. Prompts that skip any of these tend to produce vague or unpredictable output no matter how good the underlying model is.
Why these prompts use [bracketed placeholders]
You'll notice most prompts in this library include a bracketed placeholder like [topic] or [paste text] rather than being fully generic, ready-to-run text. This is deliberate: the single biggest difference between a prompt that produces a useful, specific answer and one that produces a generic, forgettable one is almost always the specific context the user supplies, not the wording of the instruction itself. These prompts are templates built around good structure — the placeholder is where your specific situation should go.
What's in each category
- Writing — editing, tightening, restructuring, and explaining, for anyone working with text regularly.
- Coding — debugging, testing, refactoring, and explaining code, framed to get explanations rather than just fixed code dropped in with no reasoning shown.
- Business — the recurring communication and analysis tasks that come up constantly in day-to-day work: follow-ups, meeting summaries, difficult conversations, and lightweight strategic frameworks.
- Study — active-recall-style prompts (quizzing, flashcards, explain-back-and-correct) that are structured around genuine learning rather than just asking a model to summarize something for you to passively read.
- Marketing — copy variations, feature-to-benefit translation, and repurposing content across formats.
Adapting a prompt instead of using it verbatim
Treat every prompt here as a first draft of your own prompt, not a finished product. Common adaptations worth making: adding a length or format constraint the base prompt doesn't specify, adding an example of the output style you want (a single well-chosen example is often more effective than a paragraph of description), and narrowing an overly broad placeholder into something specific — "topic" becomes "the return-to-office policy change" rather than staying generic.
Building your own library over time
The prompts here are a starting point, not a ceiling. As you notice yourself writing a similar prompt more than twice, that's usually a sign it's worth saving in your own running document — whether that's a plain notes file, a shared team doc, or a dedicated prompt-management tool. The habit of saving and refining reusable prompts compounds: each one you keep and improve becomes slightly better than the version before it, and over time your personal library ends up more tailored to your actual recurring work than any general-purpose list, including this one, ever could be.
Frequently asked questions
Are these prompts optimized for a specific AI model?
No — they're written to be broadly compatible across ChatGPT, Claude, Gemini, and similar assistants, since the core techniques (clear task, format, context) matter more than model-specific tricks for the vast majority of everyday use cases. Very advanced or model-specific prompting techniques are out of scope for this general-purpose library.
Why don't these prompts include a specific persona instruction like "act as an expert"?
Because independent testing on modern models has found "act as an expert" framing has a much smaller effect on output quality than it used to, and the prompt is often better spent on concrete task specification instead — the exact output format, length, and constraints you need. A few prompts here do specify a role where it genuinely helps set tone (like the audience framing in the marketing category).
Can I suggest a prompt to add to this library?
Yes — contact us with the prompt and what task it's designed for.
Is anything I search for in this library sent anywhere?
No. Search and filtering happen entirely in your browser against the prompts already loaded on the page — there's no server-side search or logging involved.