40 Best Claude Prompts (2026): Copy & Paste for Instant Results

Claude is not just another chatbot. Built by Anthropic and now running on Opus 4.6 and Sonnet 4.6, it has a 1 million token context window, extended thinking for deep reasoning, and a writing quality that consistently outperforms competitors on nuanced, long-form tasks. But like any AI tool, you get out what you put in. The prompts below are built around what Claude is genuinely strong at — not generic fill-in-the-blank templates, but specific, tested approaches that produce noticeably better output.

Before diving in: Claude Sonnet 4.6 is the right model for most of these prompts. It handles 90% of tasks at a fraction of the cost of Opus. For anything involving very long documents, complex multi-step reasoning, or large codebase analysis, switch to Opus 4.6.

Understanding What Makes Claude Different

Claude has three strengths worth understanding before you start prompting:

Massive context window. At 1 million tokens, Claude can hold an entire codebase, a full book, or months of meeting notes in a single conversation. This changes what is possible — you can ask it to analyse an entire document at once rather than feeding it in chunks.

Extended thinking. Claude can reason step by step through complex problems before giving its final answer. For hard analytical tasks, enabling this produces dramatically better results than a standard prompt.

Writing quality. Claude consistently produces more natural, nuanced prose than most models. For anything that will be read by humans — articles, emails, reports, documentation — Claude's output requires less editing.

Writing Prompts

1. Long-form article with a specific angle

Write a 1,200-word article titled "[your title]" for [target audience]. The angle is [specific perspective — not just "overview" but an actual argument or insight]. Use short paragraphs, no bullet points in the body, and a conversational but authoritative tone. End with a clear takeaway, not a generic summary.

2. Rewrite for a different audience

Here is a piece of writing: [paste text]. Rewrite it for [new audience — e.g. "a non-technical executive who has 2 minutes to read this"]. Keep the core argument intact but adjust the vocabulary, examples, and level of detail appropriately. Do not add a preamble explaining what you are about to do — just write the rewritten version.

3. Improve a rough draft

Here is a rough draft: [paste draft]. Do not rewrite it completely. Instead: (1) fix any sentences that are unclear or awkward, (2) cut any repetition, (3) strengthen the opening and closing sentences of each paragraph. Show me the improved version with your changes tracked using [square brackets] to mark what you changed and why.

4. Write in a specific voice

Write [content type] in the style of [description of voice — e.g. "a senior consultant who is direct, slightly dry, and never uses corporate jargon"]. Here is a sample of the voice I want: [paste 2-3 sentences as an example]. The topic is [topic]. Length: [word count].

5. Email that gets a response

Write a [cold / follow-up / escalation] email to [recipient role] about [topic]. My goal is [specific outcome — e.g. "a 20-minute call this week"]. Key context: [1-2 sentences of relevant background]. Keep it under 120 words. No subject line yet — I will write that separately.

6. Newsletter section

Write a 200-word section for a newsletter aimed at [audience]. The topic is [topic]. Tone: [e.g. "informed but conversational, like a smart friend explaining something they read"]. Include one concrete example or data point. Do not use bullet points.

7. Product description that sells

Write a product description for [product name]. Key features: [list them]. Target buyer: [describe them — job, problem they are solving, what they care about]. Tone: [e.g. confident and specific, not salesy]. Length: 80-100 words. Lead with the benefit, not the feature.

Research and Analysis Prompts

8. Summarise a long document

[Paste document]. Summarise this in three parts: (1) a one-sentence bottom line up front, (2) the three most important findings or arguments in plain language, (3) anything that seems uncertain, contested, or missing. Keep the whole summary under 250 words.

9. Extract structured data from unstructured text

Here is a block of text: [paste text]. Extract the following information and present it as a table: [list the fields you want — e.g. company name, date mentioned, dollar amount, key claim]. If a field is not present for a given item, write "not stated" rather than leaving it blank.

10. Compare two options

Compare [Option A] and [Option B] for someone who needs to [use case]. Cover: (1) where each one is clearly stronger, (2) where they are roughly equal, (3) which you would recommend for this use case and why. Be direct — do not hedge with "it depends" unless there is a genuine decision fork worth explaining.

11. Devil's advocate analysis

Here is an argument or plan: [paste it]. Play devil's advocate. What are the three strongest objections someone could make to this? For each objection, explain the reasoning behind it as if you genuinely believed it, then suggest how the original argument could be strengthened to address it.

12. Competitive analysis

I am researching [company or product]. Based on publicly available information, give me: (1) their apparent positioning and target customer, (2) their likely strengths based on what they emphasise, (3) their likely weaknesses or gaps based on what they do not address, (4) one thing they do that I should pay attention to. Be specific, not generic.

13. Synthesise multiple sources

[Paste source 1]. [Paste source 2]. [Paste source 3]. Synthesise these three sources into a single coherent summary. Where they agree, state the consensus. Where they conflict, note the disagreement and explain what might account for the different conclusions. Do not just summarise each one separately.

14. Build a research brief

I need to understand [topic] well enough to [specific goal — e.g. "brief a team", "make a decision", "write an article"]. Give me: (1) the core concept explained simply, (2) the 3-4 things most people get wrong or misunderstand about it, (3) the key questions I should be able to answer before I proceed, (4) 2-3 specific things worth reading or looking into further.

Coding Prompts

15. Debug with context

Here is my code: [paste code]. When I run it with input [specific input], I get [error or unexpected output] instead of [expected output]. Walk through the logic step by step, identify where it breaks down, then give me the corrected version with a brief explanation of what was wrong.

16. Code review

Review this code: [paste code]. Focus on: (1) any bugs or edge cases I have not handled, (2) performance issues for large inputs, (3) anything that would confuse another developer reading this for the first time. Do not rewrite it entirely — give specific, targeted feedback with line references where possible.

17. Write a function with tests

Write a [language] function that [describe what it does]. Requirements: [list specific requirements — inputs, outputs, edge cases to handle]. After the function, write unit tests covering the main cases including at least one edge case. Use [testing framework if relevant].

18. Explain unfamiliar code

Explain this code to me as if I understand [language] basics but am not familiar with [the specific pattern or library being used]: [paste code]. Walk through what it does in plain language, explain any non-obvious techniques, and tell me when I would use this pattern versus alternatives.

19. Refactor for readability

Refactor this code to be more readable: [paste code]. Keep the logic identical — do not change what it does. Focus on: naming, reducing nesting, splitting long functions, and adding comments only where the intent is genuinely non-obvious. Show the before and after.

20. Generate boilerplate

Generate boilerplate for [type of project or file — e.g. "a REST API endpoint in Express.js that handles file uploads"]. Include error handling, input validation, and comments marking where I need to fill in business logic. Use [framework / style conventions if relevant].

Business and Strategy Prompts

21. Stress-test a business idea

Here is a business idea: [describe it in 2-3 sentences]. Stress-test it by identifying: (1) the single assumption that, if wrong, would kill the idea, (2) the most likely reason it would fail in the first 12 months, (3) who the most dangerous competitor would be and why, (4) what would need to be true for this to be a genuinely great business. Be specific and realistic, not encouraging.

22. Write a one-page strategy brief

Write a one-page strategy brief for [initiative or project]. Structure: situation (2-3 sentences on context and why this matters now), objective (one clear sentence), approach (3-4 bullet points on how), success metrics (2-3 measurable outcomes), risks (2 main risks and mitigations). Tone: executive-ready, no fluff.

23. Meeting preparation

I have a [type of meeting — e.g. "negotiation / performance review / investor pitch"] with [who] about [topic]. My goal is [specific outcome]. Likely points of tension or difficulty: [list 1-2]. Give me: (1) the three most important things to establish early in the meeting, (2) two questions I should ask them, (3) one thing I should avoid saying or doing.

24. Job description

Write a job description for a [role title] at a [company type — e.g. "Series B SaaS startup"]. The team they will work with: [describe]. The three most important things this person will actually do day-to-day: [list them]. Tone: direct and honest, not generic corporate. Avoid buzzwords like "passionate", "rockstar", or "dynamic". Include a realistic "you might not be a fit if" section.

25. Performance review feedback

Write performance review feedback for [person's role]. Their main achievement this period: [describe]. One area for development: [describe]. Tone: honest and constructive, not padded with empty praise. Format: 3 paragraphs — strengths, development area, overall summary. Avoid vague phrases like "great attitude" or "team player" without specific evidence.

Summarisation and Document Prompts

26. Executive summary from a long report

[Paste report or long document]. Write an executive summary in under 200 words. Format: (1) what this document is about and why it matters, (2) the 3 key findings or recommendations, (3) the one thing a decision-maker absolutely needs to know. Use plain language — no jargon from the source document.

27. Meeting notes to action items

Here are raw meeting notes: [paste notes]. Extract: (1) decisions made, (2) action items with owner and deadline where mentioned, (3) open questions that were not resolved. Format as a clean table. If an owner or deadline was not stated, write "TBC".

28. Legal or contract plain-English summary

Here is a contract or legal document: [paste text]. Summarise it in plain English for someone with no legal background. Cover: what this agreement is for, the main obligations of each party, any terms that are unusual or worth flagging, and anything the reader should clarify before signing. Note: this is not legal advice.

29. Translate jargon

Here is a technical or industry-specific document: [paste text]. Rewrite it for [target audience — e.g. "a CEO with no technical background"]. Replace all jargon with plain equivalents. If a technical term is genuinely important to keep, explain it in parentheses the first time it appears.

30. Create a FAQ from a document

Here is a document: [paste text]. Generate a FAQ of 8-10 questions that someone reading this for the first time would likely ask. Write both the questions and the answers based solely on what is in the document. If a reasonable question cannot be answered from the document, include it and note "not addressed in the document."

Creative and Ideation Prompts

31. Brainstorm with constraints

Generate 10 ideas for [goal or problem]. Constraints: [list real constraints — budget, audience, timeline, what you have already tried]. For each idea, give it a one-line description and a one-line honest assessment of its main weakness. Do not give me obvious or generic ideas — push for ones that are unexpected or underexplored.

32. Name generation

Generate 15 name options for [product / company / feature]. Criteria: [list what matters — e.g. "easy to spell, not already a common word, works in English and Spanish, sounds credible not cute"]. For each name, include a one-sentence explanation of why it fits. Then flag your top 3 and explain why.

33. Tagline options

Write 10 tagline options for [product or company]. What it does: [describe]. Target audience: [describe]. Tone direction: [e.g. "confident and minimal, not clever or punny"]. For each tagline, write it out and note the emotion or idea it is trying to convey. Do not use rhymes unless they are genuinely strong.

34. Scenario planning

I am planning for [situation or decision]. Walk me through three scenarios: (1) the optimistic but realistic case, (2) the base case, (3) the pessimistic but plausible case. For each scenario, describe what conditions would lead to it and what I should do differently in preparation. Be specific about the conditions — not just "if things go well."

Prompts That Use Claude's Long Context Window

35. Analyse an entire codebase

[Paste the full codebase or relevant files]. Give me: (1) a high-level summary of what this codebase does and how it is structured, (2) the three most significant technical debt issues, (3) any security concerns I should address before shipping, (4) one architectural decision that looks like it will cause problems at scale.

36. Cross-reference multiple documents

[Paste Document A]. [Paste Document B]. [Paste Document C]. These documents cover the same topic from different sources / time periods / perspectives. Identify: (1) where they are consistent, (2) where they contradict each other and what might explain the difference, (3) what important question none of them fully answers.

37. Audit a long piece of writing

[Paste a full report, article, or document]. Audit this for: (1) logical gaps or unsupported claims, (2) internal contradictions, (3) sections that could be cut without losing anything important, (4) the single strongest section and the single weakest. Be direct — I want honest critique, not encouragement.

38. Build a knowledge base summary

[Paste all relevant documents, notes, or transcripts on a topic]. I am building a knowledge base on [topic]. Synthesise everything here into: (1) a structured overview of the topic, (2) key facts and figures worth remembering, (3) open questions or gaps in the current knowledge, (4) a glossary of important terms used across these documents.

Extended Thinking Prompts

These prompts are designed to trigger Claude's step-by-step reasoning. Start your prompt with "Think through this carefully before answering:" to activate extended thinking on Opus 4.6 or Sonnet 4.6.

39. Hard decision analysis

Think through this carefully before answering. I need to decide between [Option A] and [Option B]. Here is the full context: [describe the situation, constraints, and what you value most in the outcome]. Work through the decision step by step — consider second-order effects, what information I might be missing, and what I might be overweighting emotionally. Then give me a clear recommendation with your reasoning.

40. Complex problem solving

Think through this carefully before answering. Problem: [describe the problem clearly]. What I have already tried: [list approaches]. Constraints: [list real constraints]. I am not looking for the obvious answer — I need a solution that accounts for [the specific complexity that makes this hard]. Walk through your reasoning before giving your recommendation.

Tips for Getting More from Claude in 2026

Use the full context window. Most people prompt Claude as if it has a small memory. It does not. Paste in the entire document, the full codebase, the complete email thread. Claude performs better with more context, not less.

Tell Claude who you are and why you are asking. "I am a CFO preparing a board presentation" produces a different — and better — answer than an identical question with no context. Role and purpose shape everything about the response.

Ask for the reasoning, not just the answer. Adding "walk me through your reasoning" or "explain why" to analytical prompts gives you output you can actually evaluate and push back on, rather than just accept or reject.

Be honest about constraints. Claude responds well to real constraints — budget, audience, word count, what you have already tried. Vague prompts produce vague answers. Specific constraints produce specific, usable output.

Iterate in the same conversation. Claude maintains context across the whole thread. Rather than starting a new chat when you want a variation, stay in the same conversation: "That is close but too formal — rewrite the second paragraph to be more direct." Each iteration builds on the last.

Frequently Asked Questions

What is the best Claude model to use for prompting in 2026?

For most tasks, Claude Sonnet 4.6 is the best choice — it delivers near-Opus performance at significantly lower cost and handles the majority of writing, research, and coding prompts without compromise. Switch to Claude Opus 4.6 for tasks that require deep multi-step reasoning, large codebase analysis, or very long document processing where maximum accuracy matters.

How is prompting Claude different from prompting ChatGPT?

Claude responds better to longer, more context-rich prompts. Where ChatGPT works well with short, simple instructions, Claude is designed to handle nuance and complexity — so giving it more background, more constraints, and more specific output requirements consistently produces better results. Claude also tends to be more direct about uncertainty and less likely to pad answers with unnecessary affirmations.

What is Claude's context window in 2026?

Both Claude Opus 4.6 and Sonnet 4.6 support a 1 million token context window — roughly equivalent to several full-length novels or a large codebase. Claude Haiku 4.5 supports 200,000 tokens. This is one of the largest context windows available among frontier AI models, making Claude particularly strong for tasks that require processing very long documents in one pass.

Does Claude have extended thinking / reasoning mode?

Yes. Claude Opus 4.6 and Sonnet 4.6 both support adaptive thinking, where Claude dynamically decides when and how much to reason before answering. For complex analytical tasks, you can trigger this explicitly by asking Claude to "think through this carefully before answering" or "work through this step by step." This mode produces noticeably better results on hard problems but takes longer than a standard response.

Is Claude better than ChatGPT for writing?

For long-form writing, nuanced prose, and tasks where the output will be read by humans, Claude generally produces more natural and less formulaic results than ChatGPT. For short-form content, the gap is smaller. For coding, Claude Opus 4.6 leads most benchmarks on complex reasoning tasks, though GPT-5.3 Codex holds an edge on terminal-heavy agentic workflows. Most serious users keep both tools and use each for what it does best.

Can I use these prompts on Claude's free tier?

Yes, the free tier now gives access to Sonnet 4.6 at no cost, which means all 40 prompts in this guide will work on the free plan. Free users do have usage limits, so for heavy use — long documents, repeated analysis, or high-volume work — Claude Pro at $20/month removes those limits and adds Extended Thinking and full Claude Code access.

What is Claude best at compared to other AI tools in 2026?

Claude's three clearest advantages in 2026 are its 1 million token context window (the largest among mainstream models), its writing quality for nuanced and long-form prose, and its performance on complex coding and reasoning tasks via Opus 4.6. Where it trails competitors: it does not generate images or video natively, its voice mode is more limited than Grok or ChatGPT, and its third-party plugin ecosystem is smaller than ChatGPT's.

How do I access Claude in 2026?

Claude is available at claude.ai on web, iOS, and Android. There is also a desktop app, Claude Code for developers, and integrations including Claude in Chrome, Claude in Excel, and Claude in PowerPoint. Developers can access all models via the Anthropic API. As of March 2026, Claude Sonnet 4.6 is also available to Microsoft 365 Copilot users.