Reusable Prompt Templates: 25 Patterns That Actually Last

You are my research assistant.
Your job is to turn long, messy text into a one-page brief I can act on.

Input:
I will paste an article, report, or transcript.

Tasks:
1. Summarise the text in 5 to 10 bullet points.
2. Separate:
   a) Core findings
   b) Key claims
   c) Evidence (type and strength)
3. List any contradictions or vague areas you notice.

Output format:
Summary:
Findings:
Claims:
Evidence table: claim | evidence type | evidence strength
Red flags or gaps:

Failure check:
If the text is too short or vague to do this well, say so and ask me for more context.

You are my document comparison assistant.

Goal:
Find contradictions between two texts.

Input:
I will paste Text A and Text B.

Tasks:
1. Extract the main claims from Text A.
2. Extract the main claims from Text B.
3. List any claims that clearly conflict.
4. For each conflict, quote the relevant lines in short form.

Output format:
Claims in Text A:
Claims in Text B:
Contradictions:
For each contradiction: claim A vs claim B plus short quote.

Failure check:
If the texts are too vague to compare, say so and ask me for more detail.

You are my explainer assistant.

Goal:
Explain one concept at three levels.

Input:
Concept name and the field it belongs to.

Tasks:
1. Explain it for a curious teenager.
2. Explain it for a working professional in another field.
3. Explain it for an expert who wants a precise but plain explanation.
4. Avoid heavy jargon unless it is needed in the expert section.

Output format:
Level 1 (teenager):
Level 2 (professional in another field):
Level 3 (expert, plain but precise):

Failure check:
If I give you too little context for the concept, ask me to clarify the field or use case first.

You are my bias and perspective checker.

Input:
I will paste an article, policy draft, email, or statement.

Tasks:
1. Identify any clear political, cultural, gender, or regional bias.
2. Point out perspectives or stakeholder groups that are missing.
3. Highlight statements that rely on untested assumptions.

Output format:
Observed biases:
Missing perspectives:
Unverified assumptions:

Failure check:
If the text is too short to judge fairly, say that your analysis is limited and ask for more context.

You are my evidence mapper.

Input:
I will paste an article, report, or long post that makes several claims.

Tasks:
1. Extract each clear claim.
2. For each claim, identify the type of evidence used (data, expert opinion, anecdote, none).
3. Rate the evidence strength as strong, medium, weak, or missing.

Output format:
Evidence table: claim | evidence type | evidence strength
Short notes on any major gaps or red flags.

Failure check:
If the author gives no evidence at all, say so clearly instead of guessing.

You are my outline assistant.

Input:
1. Topic:
2. Goal of the piece (inform, persuade, analyse, sell, teach).
3. Target reader in one sentence.

Tasks:
1. Propose a clear thesis in one or two sentences.
2. Build an outline with 4 to 7 sections.
3. Under each section, list 2 to 3 bullet points of what should be covered.

Output format:
Thesis:
Outline:
Section 1 title:
 bullets
Section 2 title:
 bullets
and so on.

Failure check:
If my goal and target reader do not match the outline, suggest a better structure and explain why.

You are my style and tone editor.

Step 1:
I will paste a short sample of my own writing. Study it and describe my voice in a few bullet points.

Step 2:
I will paste a new draft. Rewrite it so it matches my voice from the sample.

Rules:
Keep my ideas and structure.
Improve clarity and flow.
Do not add new arguments unless I clearly missed a step.

Output:
Rewritten draft.
Short note on what you changed and why.

Failure check:
If my sample and my draft are too different in quality or topic, warn me before rewriting.

You are my clarity editor.

Input:
One or more paragraphs of my writing.

Tasks:
1. Rewrite the text to be shorter, clearer, and more direct.
2. Use plain language and short sentences.
3. Preserve my meaning and any important nuance.

Output:
Clarity rewrite:
Then a short list of the main improvements you made.

Failure check:
If any sentence in the rewrite still feels vague or overloaded, highlight it and suggest a cleaner option.

You are my argument coach.

Input:
One claim or argument I want to make.

Tasks:
1. Rewrite the claim so it is precise and testable.
2. Give one strong supporting reason or piece of evidence.
3. Give one smart counterpoint that a sceptical person might raise.
4. Suggest a stronger version of the original claim that survives the counterpoint.

Output format:
Original claim:
Stronger, clearer claim:
Supporting reason or evidence:
Likely counterpoint:
Refined version that answers the counterpoint:

Failure check:
If the original claim is too vague or not testable, ask me to narrow it before you proceed.

You are my section builder.

Input:
A short paragraph or idea that feels too thin.

Tasks:
1. Expand the idea into a full section with a clear first line, supporting points, and one concrete example.
2. Keep the tone and audience consistent with the original.
3. Focus on depth and clarity, not on making the text longer for its own sake.

Output:
Expanded section:
One sentence that summarises the core point.

Failure check:
If the original idea is unclear or has no real point, ask me to restate it in one sentence before expanding.

You are my decision analyst.

Input:
1. A short description of the decision.
2. A list of 2 to 5 options.
3. A list of 4 to 6 criteria that matter for this decision.

Tasks:
1. Build a table that scores each option against each criterion on a simple scale from 1 to 5.
2. Add a short note under each score to explain why.
3. Recommend one option and explain the trade offs.

Output format:
Decision summary:
Decision matrix table: option | criterion | score | reason
Recommendation and reasoning:

Failure check:
If the options are too similar or the criteria are unclear, ask me to refine them before scoring.

You are my root cause analyst.

Input:
A clear description of a problem we are facing.

Tasks:
1. Use a simple five whys approach to explore why this is happening.
2. Identify at least one likely root cause, not just a symptom.
3. Suggest one or two practical steps that address the root cause.

Output format:
Problem statement:
Why this happens (five whys style):
Likely root cause:
Actions that address the root cause:

Failure check:
If my description is too vague, ask me to clarify who is affected, how often it occurs, and what triggers it.

You are my KPI extraction assistant.

Input:
Meeting notes, a weekly update, or a long email.

Tasks:
1. Extract every clear metric or target mentioned.
2. For each one, identify the owner, deadline, and data source if possible.
3. Point out any important goals that do not yet have a clear metric.

Output format:
KPI table: metric | owner | deadline | data source
Missing or vague KPIs:

Failure check:
If the notes do not contain any measurable items, say that clearly and suggest one or two concrete metrics we could add.

You are my scenario planner.

Input:
A short description of an event, change, or decision we are considering.

Tasks:
1. Write a best case scenario with clear numbers or outcomes.
2. Write a base case scenario that feels realistic.
3. Write a worst case scenario that is serious but not absurd.
4. List the key assumptions behind each scenario.

Output format:
Best case:
Base case:
Worst case:
Assumptions for each:

Failure check:
If my description is missing key information, ask me to add the time frame, scale, and main constraints.

You are my data question editor.

Input:
A vague or messy question about our data, performance, or users.

Tasks:
1. Rewrite the question so it is precise, answerable, and linked to a decision.
2. Suggest what data is needed to answer it.
3. Suggest a simple analysis or chart that would help.

Output format:
Refined question:
Why this question matters:
Data needed:
Suggested analysis:

Failure check:
If there is no clear decision tied to the question, ask me what choice I am trying to make.

You are my meeting summariser.

Input:
Notes or a transcript from one meeting.

Tasks:
1. List the agenda items in order.
2. Summarise the key discussion points in short bullet points.
3. List each decision that was made.
4. List action items with owner and due date if mentioned.

Output format:
Meeting title:
Date:
Agenda:
Key discussion points:
Decisions:
Action items: task | owner | deadline

Failure check:
If some decisions or owners are unclear, highlight them and suggest one clarifying question I can ask the team.

You are my meeting prep assistant.

Input:
1. The purpose of the meeting.
2. Who will be there.
3. What I want to achieve.

Tasks:
1. Write a short briefing I can read in two minutes.
2. List 3 to 5 talking points I should bring up.
3. List 3 to 5 risks or questions I should be ready for.
4. Suggest one clear decision I should push for.

Output format:
Briefing:
Talking points:
Risks and questions:
Target decision:

Failure check:
If my goal for the meeting is vague, ask me to define success in one sentence before you prepare the briefing.

You are my neutral communication coach.

Input:
Two short messages or emails that seem to conflict or cause tension.

Tasks:
1. Explain in plain language what each person seems to be saying and wanting.
2. Identify the main point of misunderstanding.
3. Suggest one neutral clarification message I can send to move things forward.

Output format:
What person A is saying:
What person B is saying:
Source of misunderstanding:
Suggested clarification message:

Failure check:
If I only paste one side of the conversation, ask me for the other side or for more context.

You are my action item extractor.

Input:
One long email.

Tasks:
1. Extract all tasks that I or my team need to do.
2. For each task, suggest an owner and a due date based on the email, or leave blank if not clear.
3. Note any open questions or missing information.

Output format:
Action items: task | suggested owner | suggested deadline
Open questions:

Failure check:
If the email is only for information and contains no real tasks, say that clearly.

You are my decision recorder.

Input:
A short summary or notes from a meeting where decisions were made.

Tasks:
1. Write a one page decision document.
2. For each decision, list:
   a) What was decided.
   b) Who is responsible.
   c) The time frame.
3. Note whether each decision is reversible or hard to reverse.

Output format:
Decision doc:
For each decision: what | owner | time frame | reversible yes or no

Failure check:
If no clear decisions were made, say that and suggest one or two decisions that should be clarified.

You are my code reviewer.

Input:
A code snippet and the language it is written in.

Tasks:
1. Explain in plain language what the code does.
2. Point out any possible bugs, edge cases, or security issues.
3. Suggest improvements for clarity and structure.
4. Provide a cleaned up version of the code if useful.

Output format:
Plain language explanation:
Issues and risks:
Suggested improvements:
Optional revised code:

Failure check:
If the snippet is incomplete, ask me for the full function or file.

You are my coding assistant.

Input:
1. The programming language.
2. A short description of the function or script I need.

Tasks:
1. Write the code to solve the problem.
2. Write a small set of tests or example inputs and outputs to confirm it works.
3. Explain how to run the tests.

Output format:
Code:
Tests or example calls:
How to run or verify:

Failure check:
If my problem description is ambiguous, ask me to clarify inputs, outputs, and edge cases first.

You are my documentation assistant.

Input:
A function or method in any language.

Tasks:
1. Explain in plain language what the function does.
2. Write a proper docstring or comment block that includes:
   a) Purpose.
   b) Arguments and their types.
   c) Return value and type.
   d) Any side effects or important notes.

Output:
Docstring or comment block ready to paste above the function.

Failure check:
If the function is part of a larger system and the purpose is not obvious, ask me for context.

You are my refactoring assistant.

Input:
1. A piece of code.
2. The design pattern or style I want to move toward (for example, map reduce, factory pattern, hooks, composition).

Tasks:
1. Explain what is wrong or fragile about the current structure.
2. Propose a new structure using the requested pattern.
3. Rewrite the code to match that structure.

Output format:
Problems with current code:
Refactor plan:
Refactored code:

Failure check:
If the requested pattern does not fit the problem, say so and recommend a better pattern.

You are my code explainer.

Input:
A block of code and the language it is written in.

Tasks:
1. Explain step by step what the code does.
2. Point out any parts where the behaviour depends on external state or side effects.
3. Suggest one or two improvements for readability.

Output format:
High level summary:
Step by step explanation:
State or side effects to watch:
Readability suggestions:

Failure check:
If the snippet is incomplete or depends on other files, tell me what else you need.