Home › AI Tools & Productivity › ChatGPT & Claude Prompts
100 Original Prompt Patterns · 50 ChatGPT + 50 Claude

100 Secret ChatGPT & Claude Prompts That Feel Like Cheat Codes

These are not hidden system commands. They are memorable instruction patterns that make it easier to ask for sharper plans, better writing, deeper research, stronger study tools, smarter business analysis, cleaner code and more useful everyday answers.

Ready-to-use templatesEvergreen labelsNo fake hidden-command claims

First: What “Cheat Code” Means Here

There is no secret command console hiding inside ChatGPT or Claude. In this guide, a “cheat code” means a repeatable natural-language instruction pattern that quickly changes how the AI approaches a task. The slash labels — such as /eisenhower or /quotefirst — are mnemonic shortcuts for you, not built-in product commands.

Good prompting is less about magic wording and more about giving the model a clear role, useful context, a specific task, constraints, examples when needed, and a defined output format. Anthropic’s current prompting guidance specifically recommends clear instructions, examples, XML structure for complex Claude prompts, and careful long-context placement.

People collaborating around a laptop in a modern professional workspace
Prompting works best when you treat AI like a collaborator that needs a clear brief, useful evidence and a defined output.

ChatGPT vs. Claude: Which Prompt List Should You Use?

ChatGPT side

Fast, practical, everyday workflows

This list emphasizes productivity, monetization ideas, business planning, content, short-form scripting, studying, writing, career tasks and day-to-day decisions. ChatGPT also supports Projects, uploaded files, web search and image tools, depending on the workflow and plan.

Claude side

Structured, source-grounded, document-heavy workflows

This list leans into Claude’s documented prompting strengths: XML tags, long-document organization, quote-first grounding, careful editing, coding review, Projects and Artifacts. For complex document prompts, Anthropic recommends putting long material above the query and structuring sources clearly.

50 ChatGPT Cheat-Code Prompts

Use the category shortcuts below to jump directly to the kind of result you need.

ChatGPT: Productivity

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #1/eisenhower

What it does: Turns a messy task list into an urgent/important decision matrix.

Ready-to-use prompt

Here is my task list: [paste list]. Sort every item into the four Eisenhower Matrix quadrants. For each task, give one verdict: Do, Schedule, Delegate, or Delete. Then tell me the 3 actions I should start today.

ChatGPT #2/timeblock

What it does: Converts priorities and available hours into a realistic day plan.

Ready-to-use prompt

My priorities are [goals/tasks]. I am available from [start time] to [end time], with these fixed commitments: [commitments]. Build a realistic time-blocked schedule with breaks, buffer time, and one fallback plan if I fall behind.

ChatGPT #3/ifthen

What it does: Creates behavior triggers that make plans easier to execute.

Ready-to-use prompt

Goal: [goal]. Obstacles I usually face: [obstacles]. Create 10 specific 'If X happens, then I will do Y' implementation intentions. Keep each action small enough to do immediately.

ChatGPT #4/nextaction

What it does: Reduces an overwhelming project to the smallest concrete next move.

Ready-to-use prompt

Project: [project]. I feel stuck because [reason]. Do not give me a full plan yet. Identify the single smallest physical next action I can complete in 10 minutes or less, then tell me what should happen immediately after it.

ChatGPT #5/batch

What it does: Groups similar work to reduce context switching.

Ready-to-use prompt

Tasks: [paste tasks]. Group them into efficient batches by tool, location, mental energy, and communication type. Recommend the best order and explain which context switches you eliminated.

ChatGPT: Money & Side Income

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #6/first100

What it does: Builds a realistic plan to earn a first small amount from current skills.

Ready-to-use prompt

My skills: [skills]. Time available: [hours/week]. Budget: [budget]. Create a no-fluff plan to earn my first $100 legally and realistically. Prioritize offers I can launch within 7 days, include where to find customers, what to charge, and the first 5 outreach messages.

ChatGPT #7/validate

What it does: Pressure-tests a business or side-hustle idea before you invest.

Ready-to-use prompt

Idea: [idea]. Target customer: [customer]. Budget: [budget]. Stress-test this idea for demand, competition, willingness to pay, startup friction, and obvious failure points. Give me a cheap validation test I can run before building anything.

ChatGPT #8/nichedown

What it does: Finds narrower, more specific markets for a broad skill.

Ready-to-use prompt

My broad service/skill is [skill]. Generate 12 narrower niches where the problem is urgent and buyers can be identified. Rank them by ease of reaching customers, competition, and pricing power. Recommend the top 3.

ChatGPT #9/offerladder

What it does: Turns one service into a structured set of offers.

Ready-to-use prompt

Business/skill: [business or skill]. Design an offer ladder with a free entry point, low-ticket offer, core offer, premium offer, and optional recurring offer. For each, specify who it is for, outcome, deliverables, and a sensible price-testing range.

ChatGPT #10/passivemap

What it does: Maps expertise into digital products without pretending income is passive overnight.

Ready-to-use prompt

My knowledge/skill is [skill]. Map it into 10 digital-product ideas such as templates, checklists, mini-courses, guides, calculators, or packs. Rank them by build time and buyer usefulness, then give me a 30-day launch plan for the best one.

ChatGPT: Business

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #11/swot360

What it does: Transforms a SWOT analysis into decisions and next steps.

Ready-to-use prompt

Business: [business]. Create a SWOT analysis, but do not stop at the list. For every Strength, Weakness, Opportunity, and Threat, add 'So what?' and one concrete 'Now what?' action.

ChatGPT #12/personagen

What it does: Creates an actionable ideal-customer profile instead of a generic persona.

Ready-to-use prompt

Offer: [offer]. Build 3 realistic ideal-customer profiles. Include situation, goals, frustrations, objections, buying triggers, alternatives they use now, words they might actually say, and the best message angle for each.

ChatGPT #13/pricingteardown

What it does: Reviews pricing through value, positioning, and testing.

Ready-to-use prompt

Current offer and price: [details]. Analyze whether the pricing matches the value, customer segment, alternatives, risk, and positioning. Suggest 3 pricing experiments and what evidence would tell me each test is working.

ChatGPT #14/premortem

What it does: Finds risks by imagining the project has already failed.

Ready-to-use prompt

Assume this project failed badly 6 months from now: [project]. Work backward and list the 12 most plausible causes. Rank them by likelihood and impact, then create a prevention or early-warning action for the top 5.

ChatGPT #15/onepager

What it does: Compresses a business idea into an executive one-page brief.

Ready-to-use prompt

Turn this information into a one-page executive brief: [paste notes]. Use these sections: Problem, Customer, Solution, Why Now, Business Model, Evidence, Key Metrics, Risks, Next 30 Days, and The Ask. Keep it concise enough for a busy decision-maker.

ChatGPT: Content Creation

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #16/repurpose

What it does: Turns one piece of content into a multi-platform content system.

Ready-to-use prompt

Source content: [paste]. Repurpose it into: 1 LinkedIn post, 1 X thread, 1 Instagram carousel outline, 1 TikTok/Reel script, 1 email, and 5 short hooks. Preserve the core idea but adapt the style to each platform.

ChatGPT #17/hookstack

What it does: Generates multiple types of opening hooks instead of tiny wording changes.

Ready-to-use prompt

Topic: [topic]. Create 15 genuinely different hooks: 3 curiosity, 3 contrarian, 3 problem-first, 3 story/confession, and 3 proof/statistic-style hooks. Avoid clickbait that the content cannot deliver.

ChatGPT #18/humanvsai

What it does: Removes repetitive, generic AI-sounding writing patterns.

Ready-to-use prompt

Rewrite this draft so it sounds natural, specific, and human: [paste]. Remove generic AI transitions, inflated claims, repetitive sentence shapes, and unnecessary summaries. Preserve the facts and meaning. Then list the 5 biggest changes you made.

ChatGPT #19/adddata

What it does: Strengthens a draft with places where evidence would improve trust.

Ready-to-use prompt

Draft: [paste]. Identify where concrete statistics, examples, comparisons, quotes, or primary sources would materially improve credibility. Do not invent evidence. Mark every item I need to verify and suggest the type of source I should look for.

ChatGPT #20/objection

What it does: Builds persuasive content around reader doubts.

Ready-to-use prompt

Offer/idea: [offer]. List the 10 strongest objections a skeptical buyer or reader could have. For each, write a concise response using evidence, trade-offs, or honest limitations instead of hype.

ChatGPT: TikTok & Short-Form

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #21/hpsc

What it does: Creates a retention-focused Hook → Problem → Solution → CTA script.

Ready-to-use prompt

Topic: [topic]. Write a [duration]-second TikTok/Reel using Hook → Problem → Solution → CTA. Give me the spoken script, on-screen text, and a simple B-roll idea for each section.

ChatGPT #22/3shook

What it does: Produces openings designed to work in the first few seconds.

Ready-to-use prompt

Topic: [topic]. Write 15 hooks that can land in 1–3 seconds. Each hook must work both as a spoken line and as on-screen text. Keep them short, clear, and curiosity-driven without misleading people.

ChatGPT #23/patterninterrupt

What it does: Opens short-form content with an unexpected but relevant moment.

Ready-to-use prompt

Topic: [topic]. Create 10 pattern-interrupt openings using surprising statements, unusual questions, visual actions, or contrast. For each, explain the first visual shot and the next line that proves the hook is relevant.

ChatGPT #24/beatsheet

What it does: Maps a short video second by second.

Ready-to-use prompt

Create a second-by-second beat sheet for a [15/30/45]-second video about [topic]. Include timestamp, voiceover, on-screen text, B-roll/action, sound cue, and the purpose of each beat.

ChatGPT #25/reverseviral

What it does: Extracts reusable structure from successful examples without copying them.

Ready-to-use prompt

Here are [5–10] successful scripts/posts: [paste]. Analyze only their reusable structure: hook type, pacing, curiosity gaps, proof, payoff, CTA, and length. Create a fresh template that captures the pattern without copying wording or ideas.

ChatGPT: Studying

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #26/feynman

What it does: Combines simple explanation with active-recall checking.

Ready-to-use prompt

Teach me [topic] using the Feynman technique. First explain it simply, then ask me 5 questions one at a time. After each answer, identify the gap in my understanding and reteach only that part.

ChatGPT #27/socratic

What it does: Tutors by questioning before explaining.

Ready-to-use prompt

Teach me [topic] Socratically. Start by asking what I already understand. Then guide me with one question at a time instead of lecturing. Only explain directly when my answer shows a specific gap.

ChatGPT #28/examgen

What it does: Turns notes into a practice exam with explanations.

Ready-to-use prompt

Use only these study materials: [paste/upload]. Create a practice exam with [number] questions using a mix of multiple choice, short answer, and application questions. Add an answer key with rationales and point out which source section each question tests.

ChatGPT #29/spacedplan

What it does: Creates a practical spaced-review calendar.

Ready-to-use prompt

I need to learn [topics] by [date]. Build a spaced-repetition plan with review dates, what to recall on each date, and quick self-test methods. Spend more review time on the topics I rate as hardest: [hard topics].

ChatGPT #30/mnemonic

What it does: Creates memorable retrieval cues for difficult facts.

Ready-to-use prompt

Facts I need to remember: [paste]. Create several mnemonic options: acronym, story, visual association, rhyme, and memory palace. Keep every mnemonic factually faithful and tell me which one is easiest to retrieve under exam pressure.

ChatGPT: Learning Skills

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #31/roadmap30

What it does: Turns a skill goal into a 30-day practice roadmap.

Ready-to-use prompt

Skill: [skill]. Current level: [level]. Time per day: [minutes]. Create a 30-day roadmap with daily practice, weekly milestones, one small project, and a simple way to measure progress. Avoid resource overload.

ChatGPT #32/eli5then15

What it does: Scaffolds the same concept from simple to advanced.

Ready-to-use prompt

Explain [topic] at three levels: first for a curious 5-year-old, then a 15-year-old, then an informed adult/professional. Keep the core idea consistent and add complexity only when needed.

ChatGPT #33/misconceptions

What it does: Surfaces beginner errors before they become habits.

Ready-to-use prompt

Topic/skill: [topic]. List the 10 most common beginner misconceptions. For each, explain why it is tempting, what is actually true, and one quick test/example that makes the correction memorable.

ChatGPT #34/projectfirst

What it does: Designs a learn-by-doing project around a new skill.

Ready-to-use prompt

I want to learn [skill]. Design one beginner-friendly project that forces me to use the core 20% of the skill. Break it into milestones, define what 'done' looks like, and give hints instead of full solutions.

ChatGPT: Writing

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #35/rewriteas

What it does: Rewrites for a defined voice without copying a living writer.

Ready-to-use prompt

Rewrite this text for this audience and voice: [describe voice traits, e.g. direct, witty, conversational, concise]. Text: [paste]. Preserve meaning and facts; change rhythm, diction, structure, and level of formality.

ChatGPT #36/tighten

What it does: Cuts length while protecting meaning.

Ready-to-use prompt

Reduce this text by [X]% without losing any essential facts, argument, instructions, or emotional nuance: [paste]. Remove repetition, throat-clearing, filler, and weak qualifiers first.

ChatGPT #37/tonedial

What it does: Shows the same message across a tone spectrum.

Ready-to-use prompt

Text: [paste]. Rewrite it in five tones from 1 to 5: very formal, professional, warm-neutral, casual, and highly conversational. Keep the underlying message identical so I can choose the right level.

ChatGPT #38/inverseprompt

What it does: Reverse-engineers the characteristics behind strong output.

Ready-to-use prompt

Analyze this example: [paste]. Infer the likely instructions, audience, structure, tone constraints, and quality criteria that could produce something similar. Then write a reusable prompt template that captures those characteristics without copying the text.

ChatGPT: Career

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #39/bulletize

What it does: Turns responsibilities into outcome-oriented résumé bullets.

Ready-to-use prompt

Raw job duties: [paste]. Convert them into 6 résumé bullets using strong action verbs, scope, tools, and measurable impact where the facts support it. Do not invent numbers; mark places where I should add a metric.

ChatGPT #40/starstory

What it does: Builds structured interview stories from raw experiences.

Ready-to-use prompt

Experience: [describe]. Turn it into a concise STAR interview story: Situation, Task, Action, Result. Highlight my specific contribution, then give a 45-second and a 90-second version.

ChatGPT #41/negotiate

What it does: Creates a professional compensation negotiation script.

Ready-to-use prompt

Offer/current salary situation: [details]. Target: [target]. Write a professional negotiation script with a confident opening, evidence-based rationale, primary ask, fallback options, and responses to 5 likely employer objections.

ChatGPT #42/linkedinhook

What it does: Improves LinkedIn positioning for clarity and search.

Ready-to-use prompt

Current headline/About section: [paste]. Rewrite it to communicate role, specialty, proof, and value clearly. Give me 5 headline options and one About section that sounds credible, not buzzword-heavy.

ChatGPT: Everyday Life

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #43/decidefast

What it does: Makes a structured decision using weighted criteria.

Ready-to-use prompt

Decision: [options]. My priorities and weights are [criteria + importance]. Build a weighted decision matrix, show the trade-offs, identify what could change the result, and give a clear recommendation.

ChatGPT #44/mealmap

What it does: Turns food constraints into a realistic weekly plan.

Ready-to-use prompt

Create a 7-day meal plan for [number] people with these constraints: budget [amount], dietary needs [needs], cooking time [minutes], foods to use/avoid [list]. Reuse ingredients to reduce waste and generate one consolidated grocery list.

ChatGPT #45/explainbill

What it does: Translates confusing bills or agreements into plain language.

Ready-to-use prompt

Explain this bill/contract in plain English: [paste/upload]. Identify what I am paying for, recurring charges, deadlines, penalties, cancellation terms, unusual clauses, and questions I should ask. Do not give legal advice; flag anything that needs a professional review.

ChatGPT: Social Media, Self-Improvement & Research

Memorable patterns you can paste into ChatGPT and customize with the bracketed details.

ChatGPT #46/contentcal

What it does: Builds a goal-based monthly social content calendar.

Ready-to-use prompt

Brand/topic: [topic]. Audience: [audience]. Goal: [goal]. Build a 30-day content calendar by platform with content pillar, format, hook, CTA, and repurposing opportunity. Balance education, proof, personality, and conversion.

ChatGPT #47/habitstack

What it does: Attaches a new behavior to an existing routine.

Ready-to-use prompt

Habit I want: [habit]. Existing routines I already do reliably: [routines]. Create 8 habit-stack options in the form 'After I [existing behavior], I will [tiny new behavior].' Make the first step take under 2 minutes.

ChatGPT #48/scamper

What it does: Uses SCAMPER to generate structured variations on an idea.

Ready-to-use prompt

Idea/product/problem: [details]. Run it through SCAMPER: Substitute, Combine, Adapt, Modify, Put to another use, Eliminate, Reverse/Rearrange. Generate at least 3 useful ideas for each category, then rank the best 5.

ChatGPT #49/steelman

What it does: Challenges a viewpoint with the strongest opposing case.

Ready-to-use prompt

My current view is: [view]. Build the strongest good-faith case against it. Use the best arguments, evidence I should seek, hidden assumptions, and conditions under which the opposing view would be right. Then give a balanced synthesis.

ChatGPT #50/sourcecheck

What it does: Creates a research verification plan without inventing citations.

Ready-to-use prompt

Claim/topic: [claim]. Tell me exactly what needs verification, which primary or authoritative source types would be strongest, what search terms to use, what evidence would falsify the claim, and how to distinguish current facts from stale ones. Do not fabricate sources.

Notebook, laptop and study materials arranged for focused learning and research
The best prompt is usually a reusable workflow: context in, constraints clarified, evidence checked, and a predictable output out.

50 Claude Cheat-Code Prompts

These prompts are deliberately different from the ChatGPT list. They use structure that is especially useful for long documents, source-grounded analysis, editing, coding and visual/interactive deliverables.

Claude: Writing & Long-Form Editing

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #1/linebyline

What it does: Edits long-form writing sentence by sentence with explicit rationale.

Ready-to-use prompt

<draft>[paste draft]</draft>
Edit the draft line by line. Return a 3-column table: Original, Revision, Reason. Preserve factual meaning, flag anything that needs verification, and finish with the 3 highest-impact global edits.

Claude #2/voicematch

What it does: Uses writing samples as style evidence for a new piece.

Ready-to-use prompt

<style_examples>
[Paste 2–3 representative writing samples]
</style_examples>
<task>Write [new piece] about [topic] for [audience]. Match recurring sentence rhythm, formality, vocabulary, and structure, but do not copy phrases from the samples.</task>

Claude #3/deepedit

What it does: Separates structural, argument, and line editing into distinct passes.

Ready-to-use prompt

<draft>[paste]</draft>
Perform 3 editing passes in order: 1) structure and flow, 2) argument/evidence and reader objections, 3) sentence-level polish. Show the important changes from each pass before presenting the final revised draft.

Claude #4/redteamdraft

What it does: Stress-tests a draft from a skeptical reader’s perspective.

Ready-to-use prompt

<draft>[paste]</draft>
Act as a skeptical, well-informed reader. Identify the 7 weakest claims, unclear transitions, hidden assumptions, or credibility problems. Then rewrite only the weakest sections while preserving the author's core position.

Claude #5/registershift

What it does: Adapts tone and register while protecting factual precision.

Ready-to-use prompt

<text>[paste]</text>
<audience>[audience]</audience>
Rewrite the text for this audience at a [formal/casual/executive/technical] register. If making the language simpler would soften or change a factual claim, flag it instead of silently changing it.

Claude #6/killdarlings

What it does: Finds beloved but unnecessary passages that weaken the whole piece.

Ready-to-use prompt

<draft>[paste]</draft>
Identify sentences or paragraphs that are elegant but redundant, overwritten, off-topic, or slowing the argument. Label each KEEP, CUT, or COMPRESS with a one-line reason, then provide a leaner version.

Claude: Coding & Architecture

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #7/rootcause

What it does: Requires a testable bug hypothesis before proposing code changes.

Ready-to-use prompt

<bug>[describe bug + error]</bug>
<code>[paste relevant code]</code>
Before proposing edits, form the 3 most plausible root-cause hypotheses and list the evidence for/against each. Choose the leading hypothesis, propose the smallest test, then recommend a fix only after that reasoning.

Claude #8/twopass

What it does: Reviews code separately for correctness and maintainability.

Ready-to-use prompt

<code>[paste]</code>
Review this in two passes. Pass 1: correctness, security, edge cases, errors, tests. Pass 2: design, readability, naming, duplication, maintainability. Rank issues Critical/High/Medium/Low and include minimal fixes.

Claude #9/ultrathinkplan

What it does: For complex coding tasks, asks for deep planning before implementation.

Ready-to-use prompt

Task: [complex coding task]. Inspect the relevant architecture and constraints first. Produce a rigorous implementation plan with files/components affected, interfaces, data flow, failure modes, tests, migration risks, and rollback strategy. Do not write production code until the plan is coherent.

Claude #10/explainthenedit

What it does: Prevents blind edits to unfamiliar or legacy code.

Ready-to-use prompt

<legacy_code>[paste code]</legacy_code>
First explain what this code does, its data flow, dependencies, invariants, and risky assumptions. Then propose the smallest safe edit for [desired change]. Explicitly list what must remain unchanged.

Claude #11/refactorwithtests

What it does: Uses tests as guardrails before structural changes.

Ready-to-use prompt

<code>[paste]</code>
Goal: [refactor goal]. First identify current observable behavior and write or propose tests that lock it down. Then refactor in small steps and map every change back to a test or requirement.

Claude #12/reviewrubric

What it does: Applies an explicit engineering rubric instead of vague code review.

Ready-to-use prompt

<code>[paste]</code>
Review against this rubric: correctness, error handling, edge cases, security, performance, SOLID/DRY where relevant, readability, testability, and operational risk. Score each 1–5, explain failures, and prioritize fixes by severity.

Claude: Research & Source Synthesis

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #13/quotefirst

What it does: Grounds a long-document answer in relevant evidence before synthesis.

Ready-to-use prompt

<documents>[paste or attach sources]</documents>
First extract only the most relevant short passages from each source and label the source. Then answer [question] using those passages as grounding. Clearly separate supported findings, disagreement, and anything the sources do not establish.

Claude #14/synthbytheme

What it does: Synthesizes across sources by idea rather than one-source-at-a-time summaries.

Ready-to-use prompt

<documents>[attach/paste sources]</documents>
Question: [question]. Synthesize by theme, not by source. For each theme show: consensus, disagreements, evidence strength, gaps, and the strongest source. End with what a careful reader can conclude and what remains uncertain.

Claude #15/hypotree

What it does: Tracks competing explanations and updates confidence as evidence arrives.

Ready-to-use prompt

<evidence>[paste sources/data]</evidence>
Build a hypothesis tree for [question]. Include at least 3 competing explanations. For each, list supporting evidence, contradicting evidence, missing tests, and a confidence level. Update the ranking after considering all evidence.

Claude #16/crosscheck

What it does: Checks each important claim against multiple supplied sources.

Ready-to-use prompt

<documents>[sources]</documents>
<claims>[claims or draft]</claims>
Cross-check every material claim against the provided documents. Mark each Supported, Partially Supported, Contradicted, or Unsupported. Cite the source location and propose a safer rewrite for weak claims.

Claude #17/docstack

What it does: Uses Claude-friendly document structure for complex multi-source tasks.

Ready-to-use prompt

<documents>
<document id='1'><source>[source name]</source><content>[document text]</content></document>
<document id='2'><source>[source name]</source><content>[document text]</content></document>
</documents>
After reading all documents, answer this query: [question]. Compare evidence across documents and state uncertainty explicitly.

Claude: Productivity & Projects

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #18/projectbrief

What it does: Creates reusable Project instructions for ongoing work.

Ready-to-use prompt

Create a reusable Claude Project brief for [project]. Include: purpose, audience, my role, Claude's role, standing rules, tone, source hierarchy, definitions, deliverable formats, things to avoid, and a short pre-delivery quality checklist.

Claude #19/weeklyops

What it does: Turns project material into a repeatable operating report.

Ready-to-use prompt

Using the files and updates in this project, create this week's operations report with: wins, metrics, changes since last report, blockers, decisions needed, upcoming deadlines, owners, and the 5 highest-priority next actions. Flag missing evidence instead of guessing.

Claude #20/planthenact

What it does: Separates plan approval from execution for multi-step work.

Ready-to-use prompt

Task: [multi-step task]. First return only a plan with stages, dependencies, risks, assumptions, and the exact outputs you intend to create. Wait for approval before executing. If information is missing, list only the missing inputs that materially affect the plan.

Claude #21/selfcheck

What it does: Adds a concrete verification rubric before final delivery.

Ready-to-use prompt

Complete this task: [task]. Before finalizing, verify the result against these criteria: [criteria]. Report any criterion you could not confidently satisfy and what evidence would be needed to verify it.

Claude: Content Creation & Artifacts

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #22/artifactpage

What it does: Turns a brief into a rendered interactive deliverable.

Ready-to-use prompt

Create a polished Artifact for [landing page/tool/one-pager] about [topic]. Audience: [audience]. Required sections/features: [list]. Prioritize clear hierarchy, accessibility, responsive layout, concise copy, and a finished visual result rather than a text-only description.

Claude #23/diagramit

What it does: Converts a process into a clear visual system map.

Ready-to-use prompt

System/process: [describe]. Create a Mermaid diagram Artifact showing actors, inputs, decisions, dependencies, feedback loops, and outputs. Then explain the 3 most important bottlenecks or failure points.

Claude #24/multiformat

What it does: Adapts one source into distinct channel-native formats.

Ready-to-use prompt

<source>[paste]</source>
Turn this into four separate deliverables: blog section, email newsletter, LinkedIn post, and short-form script. Preserve facts, but adapt opening, pacing, length, formatting, and CTA to each channel.

Claude #25/stylepreset

What it does: Encodes brand style as repeatable output constraints.

Ready-to-use prompt

Brand voice: [describe or provide examples]. Create a concise reusable style specification covering tone, sentence length, vocabulary, formatting, humor, evidence standards, forbidden phrases, and CTA style. Then apply it to this draft: [paste].

Claude: Business Strategy

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #26/memoexec

What it does: Produces an executive strategy memo with explicit assumptions and risk.

Ready-to-use prompt

<context>[business context]</context>
Write an executive strategy memo with: objective, current state, assumptions, 3 options, weighted comparison, recommendation, implementation steps, key risks, leading indicators, and what would make you reverse the decision.

Claude #27/tenkteardown

What it does: Extracts a decision-useful view from a long company/report document.

Ready-to-use prompt

<report>[upload long report/10-K equivalent]</report>
Analyze it for [decision]. Extract the core thesis, key numbers, unit economics where available, risks, management claims, contradictions, and changes over time. Reference page/section locations for every material claim.

Claude #28/optionsmatrix

What it does: Evaluates choices against weighted criteria transparently.

Ready-to-use prompt

Options: [list]. Criteria and weights: [criteria]. Build a weighted decision matrix, explain every score, run a sensitivity check on the two most uncertain weights, and recommend the option that remains strongest across plausible assumptions.

Claude #29/premortemdeep

What it does: Uses deeper failure analysis for consequential plans.

Ready-to-use prompt

Plan: [plan]. Assume it failed 12 months from now. Generate internal, market, operational, financial, legal/compliance, and execution failure modes. Rank by probability × impact, identify early-warning signals, and design mitigations for the top risks.

Claude #30/steelmanthencritique

What it does: Forces a balanced strategic view before recommendation.

Ready-to-use prompt

Proposal: [proposal]. First make the strongest possible case for it. Then make the strongest case against it. Identify which disagreements are factual vs. value-based, what evidence would resolve them, and give a balanced recommendation.

Claude: TikTok & Short-Form

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #31/hookvariants

What it does: Generates multiple short-form hooks and explains the psychological mechanism.

Ready-to-use prompt

Topic: [topic]. Generate 15 distinct short-form hooks. Beside each, label the mechanism: curiosity gap, specificity, contrast, identity, urgency, surprise, proof, or story. Reject any hook the content cannot honestly pay off.

Claude #32/scriptbeats

What it does: Creates a production-ready short-form script with precise beat timing.

Ready-to-use prompt

Write a [duration]-second script about [topic]. Return a table with timestamp, voiceover, on-screen text, visual/B-roll, transition, and retention purpose. Make the payoff arrive before the final third of the video.

Claude #33/nicheremix

What it does: Adapts a successful structure to a different market without cloning it.

Ready-to-use prompt

<winning_script>[paste]</winning_script>
New niche: [niche]. Extract the structural pattern only—hook, tension, proof, sequence, payoff, CTA—then create 3 original scripts for the new niche without reusing phrases or factual claims.

Claude #34/retentionaudit

What it does: Diagnoses likely audience drop-off points in a script.

Ready-to-use prompt

<script>[paste]</script>
Audit it for retention. Identify where viewers are most likely to scroll away and why: weak hook, slow setup, repetition, unclear stakes, delayed payoff, or generic CTA. Rewrite only the high-risk beats and show before/after.

Claude: Studying

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #35/socraticteach

What it does: Runs an adaptive tutor loop based on the learner's answers.

Ready-to-use prompt

Teach me [topic] using a Socratic loop. Ask one diagnostic question at a time, adapt the next question to my answer, and do not reveal the full explanation too early. Keep a short list of misconceptions you detect and revisit them later.

Claude #36/examfrompdf

What it does: Builds a grounded practice exam from uploaded course material.

Ready-to-use prompt

<materials>[attach lecture notes/PDF]</materials>
Use only these materials to create a [number]-question exam with difficulty levels, answer key, rationales, and source page/section references. Include recall, application, and synthesis questions.

Claude #37/conceptmap

What it does: Creates a visual map of relationships among ideas.

Ready-to-use prompt

Topic/material: [topic or upload]. Build a concept map Artifact showing major concepts, prerequisites, cause/effect links, examples, and common confusions. Then identify the 5 connections I must understand before memorizing details.

Claude #38/feynmanloop

What it does: Finds gaps in the learner's own explanation and repeats until solid.

Ready-to-use prompt

I will explain [topic] in my own words. After each explanation, identify the most important gap or inaccurate connection, ask one clarifying question, and let me try again. Continue until my explanation is accurate and simple.

Claude: Analysis & Critical Evaluation

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #39/evaluatenotsummarize

What it does: Forces evidence evaluation instead of a superficial summary.

Ready-to-use prompt

<material>[paste/upload]</material>
Do not summarize. Evaluate the strength of the evidence, logic, methodology, alternative interpretations, missing data, conflicts of interest where visible, and what findings would falsify the main conclusion.

Claude #40/confidencerated

What it does: Makes uncertainty explicit finding by finding.

Ready-to-use prompt

Analyze [question/material]. For every important finding, assign High, Medium, or Low confidence and explain why. Separate direct evidence, inference, and speculation. Tell me what new evidence would most change your confidence.

Claude #41/groundedonly

What it does: Constrains the answer to supplied evidence.

Ready-to-use prompt

<source>[paste/upload]</source>
Answer this question using only the provided source: [question]. If the source does not support a claim, say 'Not established by the provided source.' Do not fill gaps with general knowledge unless I explicitly ask.

Claude #42/chartread

What it does: Extracts and interprets information from charts or dashboards.

Ready-to-use prompt

<chart>[upload image/dashboard]</chart>
Read the chart carefully. Identify title, axes, units, legend, time range, major values, outliers, trends, and misleading visual choices. Extract the underlying values where legible, then distinguish observation from interpretation.

Claude #43/assumptionaudit

What it does: Surfaces hidden premises beneath an argument or proposal.

Ready-to-use prompt

<argument>[paste]</argument>
List the hidden assumptions this argument depends on. For each assumption, explain why it matters, whether the text provides evidence, what would invalidate it, and how the conclusion changes if it is false.

Claude: Long Documents

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #44/execsummary

What it does: Creates a priority-driven one-page brief from a long document.

Ready-to-use prompt

<document>[upload long PDF/report]</document>
My priorities are: [priorities]. Produce a one-page executive summary focused only on those priorities. Include key findings, numbers, decisions, risks, unresolved questions, and page/section references.

Claude #45/extracttable

What it does: Turns a long document into structured fields for reuse.

Ready-to-use prompt

<document>[upload]</document>
Extract these fields: [field list]. Return a clean table plus CSV-ready output. Preserve exact units and dates, cite the page/section for every row, and write 'Not found' rather than guessing.

Claude #46/reformat

What it does: Converts a long report into a new structured deliverable.

Ready-to-use prompt

<report>[upload/paste]</report>
Reformat it into a [brief/presentation outline/checklist/FAQ] for [audience]. Preserve factual content and source order where it matters, remove repetition, and identify any information that would be lost in the new format.

Claude #47/citemap

What it does: Creates a verifiable claim-to-source map.

Ready-to-use prompt

<document>[upload]</document>
Create a concise summary where every material claim is followed by its page/section reference. Then add a claim map with columns: Claim, Evidence Location, Evidence Type, Confidence, and Caveat.

Claude: Brainstorming

These patterns lean into structured instructions, long-document grounding, careful editing, coding and Artifact-style deliverables.

Claude #48/divergethenconverge

What it does: Separates idea generation from selection so the first ideas do not dominate.

Ready-to-use prompt

Problem: [problem]. Phase 1: generate 30 diverse ideas without judging them. Phase 2: group them into themes. Phase 3: score the strongest 10 against [criteria]. Phase 4: recommend the top 3 and explain what makes each distinct.

Claude #49/analogyengine

What it does: Imports useful structures from other domains.

Ready-to-use prompt

Problem: [problem]. Find 8 analogous systems from unrelated domains such as biology, logistics, games, marketplaces, education, or manufacturing. Explain the mapping for each and extract one actionable idea that could transfer back to my problem.

Claude #50/constraintflip

What it does: Turns limitations into creative design prompts.

Ready-to-use prompt

Goal: [goal]. Constraints: [constraints]. For each constraint, reframe it as a creative advantage or design rule. Generate at least 3 solutions that only become possible because the constraint exists, then rank them by practicality.

7 Rules That Make These Prompts Work Better

1. Replace the brackets.

Specific inputs beat vague requests. Add your real audience, deadline, constraints, examples and desired output.

2. Give source material when facts matter.

Upload the document, paste the data, or ask for web research rather than expecting the model to guess current facts.

3. Define the decision criteria.

“Best” is meaningless until you say what matters: cost, speed, quality, risk, audience, learning value or something else.

4. Ask for uncertainty.

For research, strategy and analysis, ask the AI to separate facts, inferences, unsupported claims and what would change its conclusion.

5. Iterate on the output.

Treat the first answer as a draft. Ask what is weak, what is missing, and how to improve it against your real goal.

6. Do not confuse confidence with correctness.

AI can sound certain while being wrong. Verify important financial, legal, medical, technical and current-information claims.

7. Protect sensitive information.

Do not paste passwords, private keys, full financial credentials or confidential data unless your environment and policies explicitly allow it.

Frequently Asked Questions

Are these real hidden ChatGPT or Claude commands?

No. The slash labels are memorable names for prompt patterns. You can paste the full prompt after each label; the AI responds to the natural-language instruction, not to a secret command system.

Do I need to type the slash label?

No. The label is only a shortcut for remembering the workflow. The ready-to-use text underneath is the important part.

Can I use a ChatGPT prompt in Claude or a Claude prompt in ChatGPT?

Usually yes. Most are normal language instructions. The Claude list is simply optimized around workflows Anthropic documents particularly well, such as XML-structured prompts, long documents and Artifacts.

Why do some prompts include XML tags?

Anthropic recommends XML tags as a way to separate instructions, context, examples and documents in complex prompts. They are organizational markers, not programming requirements.

Will these prompts always produce the same answer?

No. Output can vary with model, context, files, account features and the details you provide. These patterns improve the brief; they do not guarantee a specific result.

What is the best prompt on this list?

The one tied to a real task and filled with real context. For research, start with /sourcecheck, /quotefirst or /groundedonly. For action, try /nextaction, /timeblock or /planthenact.

The Real “Cheat Code”

A clever label gets attention, but the real advantage is reusable structure. Give the model relevant context, define the task, explain constraints, specify the output, ground important claims, and improve the result through iteration. That skill stays useful even when models, features and interfaces change.

Scroll to Top