Prompt Engineering · Track 2
20 Claude Prompts You Can Copy and Use Right Now
Stop reading about it. Build something.
These are the prompt patterns that consistently produce high-quality output from Claude — across writing, coding, analysis, research, and strategy. Each includes the template and an explanation of why it works.
Try every prompt in the live ChatGPT sandbox — no copy-paste needed
The best Claude prompts follow a consistent pattern: define a role, specify the output format, set constraints, and include 1-2 examples. Claude follows structured instructions with high fidelity, so prompts that work well with Claude tend to produce reliable results every time.
Writing
Writing prompts
1
Voice-matching editor
Here are three examples of my writing: [paste examples]. Now rewrite the following draft so it matches my voice exactly — same sentence rhythm, vocabulary level, and tone. Do not add em dashes, clichés, or a summary paragraph at the end.
Draft: [paste draft]Why it works: Prevents Claude from defaulting to generic AI prose. The examples anchor the output to your actual style.
2
Long-form outline
I'm writing a [article/report/essay] on [topic] for [audience]. Create a detailed outline with: a working headline, 5-7 section headings, 3-4 bullet points per section with the key point each should make, and a note on what evidence or examples to include.Why it works: Structured output request forces Claude to organize the content before writing it — saves significant revision time.
3
Email with a clear ask
Write a [tone: professional/friendly/direct] email to [recipient] asking them to [action]. Context: [1-2 sentences of background]. The email should be under 150 words. End with a specific CTA, not an open-ended question.Why it works: The word limit and CTA constraint prevent the vague, padded emails Claude produces by default.
4
Content repurpose
I have this [blog post / transcript / report]: [paste content]. Repurpose it into: (1) a LinkedIn post under 200 words, (2) three tweet-length takeaways, (3) a TL;DR paragraph for email. Match the original's tone.Why it works: One source, three formats — saves an hour of reformatting per piece.
Coding
Coding prompts
5
Code review with criteria
Review this [language] code for: security issues (injection, auth, data exposure), performance bottlenecks, missing error handling, TypeScript type errors, and anything that would fail a production code review. Be specific about line numbers and what to fix.
[paste code]Why it works: Explicit criteria prevent vague 'looks good' reviews. Claude will surface real issues when you tell it what to look for.
6
Bug hunt with reproduction
I have a bug in my [framework/language] code. Here's what's happening: [describe symptom]. Here's what I expect to happen: [expected behavior]. Here's the relevant code: [paste code]. Walk through what's causing this, then give me the fix.Why it works: Expected vs actual behavior framing forces systematic diagnosis instead of guessing.
7
Function with tests
Write a [language] function that [description of what it does]. Requirements: [list requirements]. Include: input validation, error handling, JSDoc comments, and a test suite using [Jest/Pytest/etc] that covers the happy path, edge cases, and error cases.Why it works: Asking for tests alongside the function ensures Claude writes testable code — not just code that runs.
8
Refactor for readability
Refactor this code to be more readable and maintainable without changing behavior. Apply: extract repeated logic into named functions, replace magic numbers with named constants, improve variable names, add comments only where the why isn't obvious.
[paste code]Why it works: The 'without changing behavior' constraint is critical — it prevents Claude from helpfully 'improving' logic.
Analysis
Analysis prompts
9
Document summary with key decisions
Summarize this [document type] in three layers: (1) one-sentence TL;DR, (2) 5 key points with supporting evidence, (3) decisions or actions required from the reader. Be direct — no filler phrases like 'The document explores...'
[paste document]Why it works: Three-layer summary forces Claude to separate signal from noise and surface the actionable content.
10
Competitive analysis
Compare [Company A] and [Company B] across: pricing, target customer, key features, positioning, known weaknesses, and recent moves. Format as a table. End with: which one wins for [my use case] and why.Why it works: Table format with a winner forces Claude to make a judgment call rather than presenting 'both have pros and cons.'
11
Data interpretation
Here's a dataset: [paste data or describe it]. Tell me: (1) what's the most important trend or anomaly, (2) what's missing that would change the interpretation, (3) what decisions this data does and does not support. Be specific.Why it works: Asking for what the data doesn't support is the critical question — prevents overconfident conclusions.
12
Risk analysis
Analyze the risks of [decision / project / plan]: list each risk, rate it High/Medium/Low probability and High/Medium/Low impact, describe what would trigger it, and suggest one mitigation for each High-rated risk. Format as a table.Why it works: Probability × impact matrix forces prioritization instead of a flat list of every possible concern.
Research
Research prompts
13
Explain a complex topic
Explain [topic] to me as if I'm [level: a smart 12-year-old / a business person with no technical background / an expert in a related field]. Use analogies. Avoid jargon unless you define it. End with the three things I'd most likely misunderstand about this topic.Why it works: The misunderstanding section is the most valuable part — it surfaces the non-obvious gaps in most explanations.
14
Research synthesis
I'm researching [topic]. Here are my sources: [paste excerpts or notes]. Synthesize what they agree on, where they disagree and why, what questions they leave unanswered, and what the strongest current consensus seems to be.Why it works: Synthesis across sources is where Claude adds value — not just summarizing each one individually.
15
Devil's advocate
I believe [position]. Give me the strongest possible case against my position — not strawmen, but the actual best arguments. Assume the person arguing against me is smart and well-informed. End with which counter-argument I should take most seriously.Why it works: Explicitly asking for the best counterargument prevents Claude from generating token-friendly agreement.
16
Literature overview
Give me an overview of what is known about [topic] as of your knowledge cutoff. Include: the main schools of thought, the strongest evidence, the most contested questions, and key researchers or sources I should know about.Why it works: Structured research prompt with explicit knowledge-cutoff awareness prevents Claude from presenting outdated info as current.
Strategy
Strategy prompts
17
Decision framework
I need to decide between [option A] and [option B]. Here's my context: [constraints, goals, resources]. Evaluate both options against: short-term impact, long-term impact, reversibility, resource cost, and alignment with [goal]. Recommend one and explain the key trade-off I'm accepting.Why it works: The reversibility and trade-off framing are what separate useful strategic advice from generic pros/cons lists.
18
First-principles breakdown
Break [problem / challenge / question] down to first principles. What are we actually trying to achieve? What assumptions is the conventional approach making? What would a solution look like if we ignored how it's usually done?Why it works: Forces Claude to challenge the frame of the question, not just answer it as posed.
19
Pre-mortem
It's [one year from now] and [project / plan] has failed completely. What went wrong? List the five most likely causes of failure — be specific to this context, not generic. Then for each: what could we do now to prevent it?Why it works: Pre-mortem framing surfaces implementation risks that forward-looking planning systematically misses.
Go deeper with the Prompt Engineering track
Track 2 at Learn to GPT covers chain-of-thought reasoning, structured output, few-shot examples, prompt evaluation, and optimization. Every lesson has a live ChatGPT sandbox so you practice the patterns immediately — not just read about them.
Frequently Asked Questions
What makes a Claude prompt effective?
Effective Claude prompts have specificity (clear role, audience, output format), structure (organized with examples and constraints), and context (relevant background). The pattern of role + format + constraints + examples consistently produces the best results.
Can I use the same prompts for Claude and ChatGPT?
Many prompts work across both models. Claude responds especially well to structured instructions with explicit constraints. Prompts optimized for Claude tend to work well everywhere because they are more precise.
How do I write a system prompt for Claude?
Start with a role definition, then specify the output format, add constraints (word count, tone, what to avoid), and include 1-2 examples of ideal output. Claude follows structured system prompts with high fidelity.
What are the most common prompt mistakes?
Being too vague (no role, format, or constraints), providing no context (Claude cannot infer your industry or goals), and asking for everything at once. Breaking complex tasks into steps produces better output.
Practice every prompt in the live sandbox
Free forever. No credit card. Start in 60 seconds.