Prompt Engineering Cheat Sheet

Anatomy of a good prompt

Strong prompts tend to include four ingredients. You will not need all four every time, but when a result disappoints, one of these is usually missing.

• Task — the verb. Summarize, compare, rewrite, list, draft. Be specific about the action you want.
• Context — what the model needs to know: the audience, the situation, the source material.
• Constraints — length, tone, format, what to include or leave out.
• Example — when format matters, show one sample of the output you want.

New to prompting? Start with Prompt Engineering Basics, then keep this page bookmarked.

Quick checklist before you hit enter

• Did I name the specific task, not just the topic?
• Did I say who the output is for?
• Did I set a length and format?
• Did I paste in the actual text or data, instead of assuming the model knows it?
• Did I ask for one task, not three bundled together?

Copy-paste templates

Replace the bracketed parts with your details.

Summarize

Summarize the text below in [5 bullet points] for [a busy manager].
Keep only the decisions and action items.

[paste text]

Explain

Explain [topic] to [a complete beginner]. Use one everyday analogy,
avoid jargon, and keep it under [150 words].

Compare

Compare [option A] and [option B] for [my situation].
Cover cost, ease of use, and risks, then end with a recommendation
and the single strongest reason for it.

Critique

Review the draft below as [a skeptical editor]. List the 3 biggest
weaknesses, quote the sentence where each appears, and suggest a fix.

[paste draft]

Brainstorm

Give me [10] ideas for [goal]. Make the first 5 safe and practical
and the last 5 unconventional. One line each, no explanations.

Rewrite

Rewrite the text below for [audience] in a [friendly professional] tone.
Keep the meaning and all facts identical. Target length: [100 words].

[paste text]

Draft

Draft a [short email] to [recipient] that [goal].
Tone: [polite but direct]. Include [key point] and end with
a clear next step.

Fixes for common failures

• Output too generic — add context and constraints. "Write about productivity" invites filler; "list 5 productivity tactics for a remote developer who hates meetings" does not.
• Wrong format — describe the format explicitly, or show a one-line example of it.
• Too long or too short — give a hard number: words, bullets, or paragraphs.
• Confidently wrong facts — ask it to flag anything it is unsure about, and always verify names, numbers, and quotes yourself. See How to Verify AI Answers.
• Ignored part of your request — you probably asked for several things at once. Split the job into separate prompts.
• Good but not quite right — do not start over. Reply with a correction: "Closer. Now make it shorter and drop the second point."

Level up

Templates get you speed; patterns get you range. When you are ready for reusable techniques like role prompting and few-shot examples, move on to the Prompt Patterns Cheat Sheet or the CLEAR Prompt Framework.