These templates are designed to be copied into ChatGPT (or any LLM) and filled in with your details. They work because they enforce a simple structure: goal, context, constraints, and format. If you’re new to this, read The Prompt Recipe first—it explains why these patterns work.
Related: questions to ask ChatGPT · reduce hallucinations · prompt lists · funny prompts · deep prompts
Treat each template as a starting point. Replace the bracketed parts, then add any constraints that matter to you (time, budget, audience, tone, tools, or what to avoid). If the first answer isn’t right, don’t start over—iterate: ask for 3 alternatives, request a critique, then ask for a revised version.
Want ready-made prompts without templates? Browse all prompt lists or get inspiration from the daily archive.
Planning templates are best when they produce a concrete artifact: a checklist, timeline, meeting agenda, or risk list. If you want more, see work & productivity prompts.
Act as a project manager.
Goal: [deliverable].
Context: [team / situation / current state].
Constraints: [deadline], [budget], [tools], [what to avoid].
Output:
1) milestones
2) weekly plan
3) risks + mitigations
4) success metrics (KPIs).Plan my week around these priorities: [list].
Context: I have [X] hours/day, meetings on [days].
Constraints: include breaks, and keep tasks to 45–90 minutes.
Output: a calendar-style schedule + a short “must-do” list.Create a meeting agenda for [topic].
Goal: [decision / alignment / brainstorm].
Context: attendees are [roles].
Constraints: 30 minutes total.
Output: timed agenda + prep questions + a follow-up email draft.Writing templates work best when you define audience + tone + structure. For more depth, see Prompting for Writing.
Write a concise email to [recipient] about [topic].
Audience: [who reads it]. Tone: [friendly / direct / calm].
Constraints: under [X] words. Avoid jargon.
Output: email + 3 alternative subject lines.Turn these notes into a clear outline:
Notes: [paste]
Audience: [who]. Goal: [what they should understand/do].
Format: H2 headings + bullets. End with a 5-bullet summary.Rewrite this to be 15% shorter and clearer.
Keep key terms unchanged: [list terms]
Text: [paste]The best learning prompts ask the model to teach, then test you. See Prompting for Learning for more.
Teach me [topic] step-by-step:
1) simple explanation (no jargon)
2) a worked example
3) a 5-question quiz (with answers hidden until I respond)
4) corrections + next practice tasks.Create 20 flashcards about [topic].
Include 5 misconception cards (common wrong ideas).
Format: Q/A pairs.I’m studying [topic]. Make a 2-week plan.
Context: I can study [minutes] per day.
Constraints: I learn best by doing; include mini-projects.
Output: daily tasks + weekly review checklist.For code, include environment details and ask for tests. For a deeper guide, see Prompting for Coding.
Act as a senior engineer.
Task: [feature / bug fix].
Environment: [language], [framework], [versions], [OS].
Constraints: [performance/security], [no new deps], [style].
Output:
1) approach
2) code changes (files + snippets)
3) tests
4) edge cases.Debug this error.
Error: [paste stacktrace]
Code: [paste minimal code]
What I tried: [list]
Output: likely causes + the smallest fix + a test that prevents regression.Review this code for bugs, security, and readability.
Then propose a safer refactor that keeps behavior the same.
Code: [paste]These templates are for summarizing your own input (notes, transcripts, docs). When accuracy matters, constrain the model to your text. If you’re tackling factual topics, pair this with Reduce Hallucinations.
Use only the information in the text below.
If something is missing, say “unknown”.
Then produce:
- a 10-bullet summary
- key terms + definitions
- open questions
Text: [paste]Summarize this into a 1-page memo.
Audience: [team/manager/client]
Format:
1) context
2) key points
3) recommendations
4) risks / unknowns
Text: [paste]Extract structured data from this text.
Return JSON with fields: [list fields].
Text: [paste]Decision templates work when you provide criteria and ask for trade-offs. If you like “thinking” prompts, browse deep prompts.
I’m choosing between [A], [B], [C].
My priorities: [ranked list].
Constraints: [budget/time/skills].
Output:
- decision matrix
- recommendation + reasoning
- biggest risks + mitigations
- what info would change your mind.Help me decide: [question].
First ask me 5 clarifying questions.
Then give 3 options and recommend one.
Output: a short plan + “next 3 actions”.Turn this messy situation into a decision.
Situation: [paste]
Output:
1) what decision needs to be made
2) options
3) criteria
4) recommended choice.Creativity templates work best when you demand variety. For playful prompts, see funny prompts.
Give me 12 ideas for [thing].
Make them diverse:
- 3 practical
- 3 weird
- 3 emotionally resonant
- 3 funny
Then pick the best 2 and explain why.Generate 20 “what if” questions about [topic].
Mix: science, philosophy, everyday life, and humor.Write 5 short prompts that test an AI’s reasoning about [topic].
Each prompt should include a tricky constraint or edge case.If you want prompts explicitly designed to probe model behavior, try LLM limit tests.
When you care about correctness, append one or more of these lines. They push the model to be explicit about uncertainty and verification.
Before answering, list assumptions you're making.
Label confidence (high/medium/low) per claim.
Then include a “verification checklist”.Give 3 alternative answers and explain when each is appropriate.List 10 edge cases / failure modes and how to handle them.Prompt templates are reusable patterns with placeholders that you fill in. They help you get consistent results because they force goal, context, constraints, and format.
Replace placeholders with your specifics, then add constraints that matter. If the answer is off, ask clarifying questions and request a revision.
Because the prompt is generic. Add constraints, ask for a concrete output format, and request multiple options before choosing one to refine.
Ground the model in your text, request assumptions and confidence labels, and include a verification checklist. See How to Reduce Hallucinations.
Next: Questions to Ask ChatGPT · The Prompt Recipe · All prompt lists · Suggest a prompt