Pick a category. Find a task that matches your work. Copy the prompt and paste it into claude.ai — replace the brackets with your own details. Most people get useful output in under 3 minutes.
Paste any document, report, or email thread. Ask AI to pull out the 3 most important points. You'll cut reading time by 70% and immediately see why this changes how you work.
Paste any document and get back a clear, accurate summary. Works for PDFs, emails, reports, articles — anything you can paste as text. Most people find this cuts reading time by 70%.
Summarize this document for me. I need: - The 3 most important points - Any decisions that need to be made - Key dates or deadlines mentioned - Anything I need to follow up on Document: [paste your document here]
Describe what you need to say and get a well-written draft back. Edit it before you send — AI drafts are a starting point, not a final product. Always review before hitting send.
Write a professional email for me. Context: [explain the situation — who are you writing to, what's the relationship, what happened] Goal: [what do you want the email to accomplish] Tone: [formal / friendly / direct / apologetic — pick one] Keep it under [X] lines. End with a clear next step for the reader.
Paste a raw transcript — from Zoom, Teams, or even rough notes — and get back structured meeting notes with decisions, action items, and who said what.
Turn this meeting transcript into clean notes. I need: - A 2-sentence summary of the meeting - Decisions that were made (with who decided) - Action items (with owner and deadline if mentioned) - Open questions or things that need follow-up Transcript: [paste your transcript here]
Paste 5–10 articles or news items and get back a synthesized digest that highlights what matters and why it's relevant to your work. Build a weekly habit with this one.
I'm a [your job title] in [your industry]. Here are [X] articles from this week. For each, give me: - One sentence on what happened - Why it matters to someone in my role - Whether I need to act on this or just be aware Then give me an overall "this week's theme" in 2 sentences. Articles: [paste article titles + text or summaries]
Paste what you wrote and tell Claude what you want it to sound like. It'll rewrite while keeping your meaning intact. Always read before you send — you know your audience better than AI does.
Rewrite this for me. Keep my meaning exactly the same. Make it: [more formal / clearer / shorter / more confident / easier to read — pick one or describe what you want] Audience: [who will read this — their role, how well they know the topic] Original: [paste your text here]
Whether it's a legal concept, a technical system, or a medical condition — AI can help you translate it into language your audience will actually understand and act on.
Explain [topic] to someone who has no background in [field]. They understand: [what they do know — their job, their experience] They need to understand: [what specific part matters for them] Use an analogy if it helps. Avoid jargon. Keep it under 200 words.
Tell Claude who you're presenting to, what you want them to do after, and your key points. It'll structure a clear outline with a logical flow — you fill in the details you know better than any AI.
Create a presentation outline for me. Audience: [who they are, their level of familiarity with the topic] Goal: [what you want them to do or believe after the presentation] Time limit: [X minutes] Key points I need to cover: [list them] Give me slide titles, one sentence per slide on what goes there, and a suggested opening hook.
Before your team adopts a new AI tool, there are specific security questions you need answered. This prompt gives you a structured vendor assessment checklist you can use in any evaluation.
I'm evaluating [AI tool name] for use in our organization. We work in [industry]. Give me a security and compliance assessment checklist covering: 1. Data handling — where does our data go, how is it stored, can it be used for training? 2. Access controls — how are user permissions managed? 3. Compliance — what certifications does it have (SOC 2, ISO 27001, HIPAA, GDPR)? 4. Incident response — what happens if there's a breach? 5. Integration risks — what does it need access to in our environment? Format as a checklist I can fill in with their answers.
A one-page AI usage policy tells your team what tools are approved, what data can't go in, and what requires human review. This is the first governance artifact every organization needs.
Write a one-page AI usage policy for our team. Industry: [your industry] Team size: [approximate] Current AI tools in use: [list any — Claude, ChatGPT, Copilot, etc.] Data sensitivity: [describe what kind of data your team handles] The policy should cover: - Approved AI tools and their permitted uses - Data that must NEVER go into an AI tool (be specific) - Tasks that require human review before acting on AI output - How to report concerns or incidents - Who owns the policy and when it will be reviewed Keep it under one page. Plain language — not legalese.
Prompt injection is the #1 vulnerability in LLM applications — it's when a user's input tricks the model into ignoring its instructions. This prompt walks you through a structured security test.
I'm testing [AI application name] for prompt injection vulnerabilities. The app: [describe what it does — customer support bot, document summarizer, etc.] The system prompt purpose: [what it's supposed to do and not do] Generate a test matrix for me with: 1. Direct injection attempts (user tries to override system prompt) 2. Indirect injection attempts (malicious content in data the app processes) 3. Role-play / persona bypasses 4. Instruction smuggling via special formatting For each category, give me 3 specific test inputs I can try, and describe what a "fail" looks like — i.e., what behavior means the injection worked.
Paste an abstract or full paper text and get back a plain-English summary focused on what's clinically relevant. Always verify against the source — AI can miss nuance in complex studies.
Summarize this research paper for me. I'm a [your role] in [your specialty]. Tell me: - What question the study was trying to answer - How they studied it (briefly — study type, sample size) - What they found - Whether the findings are likely to change clinical practice, and why or why not - Any major limitations I should know about Paper: [paste abstract or full text]
Write patient-facing content at the right reading level and tone. Never include actual patient data — write for the condition generally, then review and personalize yourself before sharing.
Write patient education material about [condition or procedure]. Target reading level: [6th grade / 8th grade / plain English for non-native speakers] Key things the patient needs to understand: - What this condition/procedure is - What to expect - What they need to do at home - Warning signs to watch for - When to call us Do NOT include any personal medical advice. Write for general patient education only. Length: one page or less.
Before running any process with AI in a healthcare context, you need to know if patient data is involved. This prompt helps you assess a specific workflow for compliance risk before you build anything.
I want to use AI for this healthcare workflow: [describe the process in detail] Help me assess: 1. Does this workflow involve PHI (protected health information)? What specifically? 2. Which steps could be done with AI on de-identified or general information? 3. Which steps require a human to handle because of PHI or clinical judgment? 4. What would a compliant version of this workflow look like? I'm not asking for legal advice — I want a practical risk assessment to take to my compliance team.
Paste the error message and the relevant code and Claude will identify what's going wrong and how to fix it. Works best when you include the full error, not just the last line.
I'm getting this error and I can't figure out why: Error message: [paste the full error / stack trace] Relevant code: [paste the function or section where the error happens] What I expected to happen: [describe] What I was doing when it broke: [describe] Tell me: what's causing this, and what's the fix? Explain why so I understand, not just what to change.
Paste unfamiliar code and get a plain-English explanation of what it does. Ask about specific parts. This speeds up onboarding and code review faster than anything else.
Explain this code to me. I [am familiar / am not familiar] with [language or framework]. Tell me: - What this code does overall in plain English - What each main section does - Any parts that are unusual, risky, or that I should pay attention to - Any obvious issues or improvements Code: [paste code here]
The Sentry for AI SDK adds observability to your LLM calls — tracking token usage, latency, errors, and the content of each call — in a few lines of code. Essential before going live with any AI feature.
I'm building an AI application using [Anthropic Claude / OpenAI / LangChain — pick one]. Help me add production monitoring using Sentry for AI. I need to track: - Errors and failures in my LLM calls - Latency per request - Token usage and cost - The content of inputs and outputs (for debugging) Give me the setup code and explain each step. I'm using [Python / Node.js — pick one]. My app does: [brief description of what your app does]
Describe the task you want to automate and Claude will help you design a workflow — including where AI fits, where a human needs to stay involved, and what tool to use to build it.
I want to automate this task: [describe it step by step — what triggers it, what you do, what the output is, who uses it] Help me design an automation. Tell me: 1. Which steps AI can handle (and which prompt to use) 2. Which steps need a human to review before anything happens 3. What tool to use to connect everything (n8n, Zapier, Make, or a script) 4. What could go wrong and how to handle it I have access to: [list tools — Slack, Gmail, Google Sheets, Notion, etc.] I [can / cannot] write code.
Paste document content and tell Claude what fields to pull out. It returns structured data you can paste directly into a spreadsheet. Works on contracts, invoices, applications, and more.
Extract structured data from this document. Return a table with these columns: [list the fields you want — e.g., Name, Date, Amount, Contract type, Expiry date] If a field is missing from the document, put "not found." If a field is ambiguous, flag it with a note. Document: [paste document text]
Tell Claude what you want to learn, how much time you have, and what you already know. It'll give you a practical plan with specific milestones — not just a vague list of topics to "research".
Create a practical learning plan for me. I want to learn: [specific skill or topic] My current level: [complete beginner / some familiarity / know basics, want to go deeper] Time I can commit: [X hours per week] My goal: [what I want to be able to DO when I'm done — be specific] I prefer learning by: [reading / watching videos / building projects / a mix] Give me a week-by-week plan with specific things to do each week — not just topics to "research."
Practice answering interview questions, get feedback on your answers, and find gaps in your preparation before the real thing. Works for any role — not just AI jobs.
Help me prepare for a job interview. Role I'm interviewing for: [job title and company type] My background: [brief summary — current role, years of experience, relevant skills] Format of interview: [behavioral / technical / case study / panel — or mix] Give me: 1. The 5 questions most likely to come up for this role 2. What the interviewer is really looking for with each one 3. One strong answer structure for the hardest question 4. Two things I should ask them at the end
Every prompt is drawn from real published sources and tested in practice. Click any to explore the original research and curriculum.
Prompts are the entry point. The paths below give you structured curriculum, portfolio projects, and a career direction.