30 Claude Prompts That Build a Working MVP
Describe the product and Claude returns a clickable MVP as one self-contained HTML file with localStorage state — no backend, previewable instantly. Prompts for SaaS UIs, marketplaces, booking apps, directories, AI wrappers, plus the one-page specs to plan it all.
In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 15 prompts are free instantly — no signup needed. Hand-curated and tested by the AI Academy team.
SaaS & Product MVPs
5 promptsSaaS Dashboard MVP
1/30You are a senior product engineer and product designer who ships scrappy but credible v1 SaaS apps. <context> I am a founder validating a SaaS idea. Build a working MVP as one self-contained HTML file that uses inline CSS, vanilla JavaScript, and localStorage for all state, so I can open it and click through the whole product without any backend. It must be a previewable artifact. </context> <inputs> - Product in one line: [WHAT IT DOES] - Core user: [WHO] - The one job it must do well: [PRIMARY ACTION, E.G. TRACK X, CREATE Y] - Key objects the user manages: [E.G. PROJECTS, TASKS, CLIENTS] - Brand vibe: [E.G. CLEAN B2B, PLAYFUL] - Seed data I want preloaded: [2-3 EXAMPLE RECORDS] </inputs> <task> Build a single-page SaaS app shell: a top bar with product name and a primary action button, a left sidebar nav, and a main content area showing a list of the core objects with create, edit, and delete working against localStorage. Include an empty state, a create modal with a real form, inline editing, and a small stats summary at the top (counts derived from the data). Preload the seed records so the app looks alive on first open. </task> <constraints> - One self-contained HTML file; no frameworks, no CDN dependencies except Google Fonts. - All data persists in localStorage and survives refresh; no fake buttons — every control works. - Accessible forms with labels, keyboard-closable modal, responsive down to mobile. </constraints> <format> Return the full HTML as an artifact, then list which features are real vs. stubbed and the first backend endpoint you would build to make it production-ready. </format>
Builds a clickable SaaS dashboard MVP with working CRUD on localStorage as a previewable artifact.
Pro tip: Tell Claude the single metric your stats bar should track; founders learn more from a real number than from a generic chart.
Internal-Tool MVP
2/30You are a full-stack engineer who builds fast internal tools for small teams. <context> My team needs a lightweight internal tool to replace a messy spreadsheet. Build it as one self-contained HTML file using inline CSS, vanilla JS, and localStorage so anyone can open it in a browser with no setup. It must be a previewable artifact. </context> <inputs> - What the tool manages: [E.G. SUPPORT TICKETS, INVENTORY, REQUESTS] - Fields per record: [LIST THE COLUMNS] - Statuses or stages: [E.G. NEW, IN PROGRESS, DONE] - Who uses it: [INTERNAL ROLE] - Bulk actions I need: [E.G. MARK DONE, FILTER, EXPORT] - Seed rows to preload: [2-4 EXAMPLE RECORDS] </inputs> <task> Build a table-driven internal tool: a filterable, sortable table of records with add-row, edit-in-place, delete, and a status dropdown per row; a top filter bar (search plus status filter); a counts summary by status; and an Export to CSV button that downloads the current data. Persist everything to localStorage and preload the seed rows. </task> <constraints> - One self-contained HTML file; no frameworks, Google Fonts allowed. - CSV export must produce a valid downloadable file from the live data. - Table is keyboard-navigable and readable; sensible empty state. </constraints> <format> Return the full HTML as an artifact, then explain the data shape you used and how to swap localStorage for a shared database later. </format>
Generates a filterable, exportable internal-tool MVP that replaces a spreadsheet as a previewable artifact.
Pro tip: List your exact spreadsheet column headers in the inputs so the tool maps one-to-one and your team feels at home instantly.
CRUD Product MVP
3/30You are a product engineer who builds clean, complete CRUD apps for early-stage products. <context> I need the core CRUD loop of my product working end to end. Build it as one self-contained HTML file with inline CSS, vanilla JavaScript, and localStorage persistence, returned as a previewable artifact I can click through. </context> <inputs> - The single entity my product is built around: [E.G. RECIPE, HABIT, DEAL] - Fields for that entity: [LIST FIELDS AND TYPES] - How users view the list: [CARDS / TABLE / KANBAN] - A detail view per item should show: [WHAT DETAILS] - Sort or filter I care about: [E.G. BY DATE, BY TAG] - Seed items to preload: [2-3 EXAMPLES] </inputs> <task> Build a full create-read-update-delete app for the entity: a list view in the chosen layout, a create form, a detail view, edit, and delete; client-side validation on the form; one sort and one filter control; and a small header showing the total count. All state lives in localStorage and persists across refresh. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Validate required fields and show inline errors; confirm before delete. - Responsive, accessible, with a real empty state and a clear primary CTA. </constraints> <format> Return the full HTML as an artifact, then describe the entity schema and which fields you would index if this moved to a real database. </format>
Builds a complete CRUD MVP around your core entity with validation and persistence as a previewable artifact.
Pro tip: Pick the smallest possible entity — one object done fully beats five objects half-built when you are testing demand.
Subscription-Box MVP Storefront
4/30You are an e-commerce engineer who builds DTC subscription-box storefronts. <context> I am launching a subscription box and need a working storefront MVP. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage so I can demo the full subscribe flow with no backend. Return it as a previewable artifact. </context> <inputs> - The box and what is inside: [E.G. ARTISAN COFFEE, MONTHLY] - Plans and prices: [E.G. MONTHLY $29, QUARTERLY $79] - Box-size or variant options: [E.G. SOLO, COUPLE, FAMILY] - Audience and vibe: [WHO IT IS FOR, AESTHETIC] - Proof: [REVIEWS, MEMBER COUNT] </inputs> <task> Build a subscription-box storefront: a hero with the box concept and a Choose Your Plan CTA, a plan-selector with the variant and frequency options updating a live price, a how-it-works three-step section, a what-is-inside showcase, a reviews strip, and a checkout-summary panel that stores the chosen plan in localStorage and shows a confirmation screen on subscribe. The plan picker and price must be fully interactive. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; styled color blocks as product photos. - Price updates instantly from the selected plan and variant; subscribe writes the selection to localStorage and shows a confirmation. - Responsive and accessible; real benefit copy, no lorem ipsum. </constraints> <format> Return the full HTML as an artifact, then explain the pricing-state logic and where a real payment provider would hook in. </format>
Builds an interactive subscription-box storefront MVP with a live plan picker and checkout flow as a previewable artifact.
Pro tip: Give Claude your real plan prices so the live total is believable in demos instead of placeholder numbers.
Freemium SaaS With Usage Meter
5/30You are a product engineer who designs freemium SaaS UIs with usage limits and upgrade prompts. <context> I want to validate a freemium model. Build an MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage that simulates a free tier with a usage cap and an upgrade path, previewable instantly as an artifact. </context> <inputs> - Product and core action: [WHAT THE USER DOES] - The metered unit: [E.G. CREDITS, EXPORTS, PROJECTS] - Free-tier limit: [NUMBER] - Paid tiers and prices: [LIST] - What unlocks on upgrade: [BENEFITS] </inputs> <task> Build a SaaS app shell where the user performs the core action, each use decrements a usage meter stored in localStorage, and hitting the free-tier cap triggers an upgrade modal showing the paid tiers and benefits. Include a visible usage bar, a reset-month demo button, and a simulated upgrade that lifts the cap. Make the core action and the metering genuinely work. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - The usage meter and cap persist in localStorage; the upgrade modal is keyboard-closable and accessible. - Clear, non-pushy upgrade copy that names the unlocked benefit. </constraints> <format> Return the full HTML as an artifact, then explain the metering logic and what you would log to learn where users hit the wall. </format>
Builds a freemium SaaS MVP with a working usage meter, cap, and upgrade prompt as a previewable artifact.
Pro tip: Set the free-tier cap low so your demo actually hits the paywall; that moment is the whole point of the test.
XML tags are just the start. Learn the full Claude workflow.
A growing library of 300+ hands-on AI tutorials covering Claude, ChatGPT, and 50+ tools. New tutorials added every week.
Marketplace & Directory MVPs
5 promptsTwo-Sided Marketplace MVP
6/30You are a marketplace engineer who builds two-sided platforms for early validation. <context> I am testing a marketplace idea. Build an MVP as one self-contained HTML file using inline CSS, vanilla JS, and localStorage so both sides of the market work without a backend, returned as a previewable artifact. </context> <inputs> - What is bought and sold: [E.G. FREELANCE GIGS, USED GEAR, LESSONS] - The two sides: [E.G. BUYERS AND SELLERS, CLIENTS AND PROS] - Listing fields: [TITLE, PRICE, CATEGORY, DESCRIPTION, ETC] - How buyers find listings: [SEARCH, CATEGORY, FILTERS] - Seed listings to preload: [4-6 EXAMPLES] </inputs> <task> Build a marketplace with a buyer view (browse, search, category filter, listing cards, and a listing detail) and a simple seller view (a form to post a new listing that appears immediately in the browse grid). Include a contact-seller or request action that stores an inquiry in localStorage. Preload the seed listings and persist everything across refresh. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; labeled color blocks for listing images. - Posting a listing updates the live grid; search and filter operate on the real data. - Accessible forms and cards; responsive; sensible empty states. </constraints> <format> Return the full HTML as an artifact, then explain the listing data shape and the trust and payment features you would add for a real launch. </format>
Builds a working two-sided marketplace MVP with browse, post, and inquiry flows as a previewable artifact.
Pro tip: Seed the supply side first — a marketplace with six real-looking listings demos far better than an empty grid.
Niche Directory MVP
7/30You are a product engineer who builds searchable directory and listing sites. <context> I want to launch a niche directory to test demand. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, previewable instantly as an artifact, with real search and filtering. </context> <inputs> - What the directory lists: [E.G. AI TOOLS, LOCAL ROASTERS, REMOTE JOBS] - Each entry has fields: [NAME, CATEGORY, TAGS, LINK, DESCRIPTION, RATING] - Filters that matter: [CATEGORY, TAGS, PRICE, ETC] - Sort options: [E.G. POPULAR, NEWEST, A-Z] - Seed entries to preload: [6-10 EXAMPLES] - Submit-an-entry: [YES/NO, FIELDS] </inputs> <task> Build a directory with a search bar, a sidebar of category and tag filters, a results grid of entry cards, and a sort control — all operating on the live data. Add an entry-detail view and, if requested, a Submit Listing form that adds to the directory and persists in localStorage. Show a result count and a clear empty state. Preload the seed entries. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Search, filter, and sort all work together on the real dataset; URL-free state is fine. - Accessible filters and cards; responsive; submitted entries appear immediately. </constraints> <format> Return the full HTML as an artifact, then describe the entry schema and how you would moderate and rank submissions at scale. </format>
Builds a searchable, filterable niche directory MVP with optional submissions as a previewable artifact.
Pro tip: Hand Claude one fully filled-in entry as a template so every seeded listing has consistent fields and the filters actually bite.
Booking / Appointment App MVP
8/30You are a product engineer who builds scheduling and booking apps. <context> I need a working booking MVP to validate an appointment-based service. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage so the full book-a-slot flow works with no backend, returned as a previewable artifact. </context> <inputs> - What gets booked: [E.G. HAIRCUTS, CONSULTATIONS, COURT TIME] - Services or session types: [NAME, DURATION, PRICE] - Availability rules: [DAYS, HOURS, SLOT LENGTH] - Info to collect from the customer: [NAME, EMAIL, NOTES] - Confirmation behavior: [WHAT THE USER SEES AFTER BOOKING] </inputs> <task> Build a booking app: a service picker, a calendar or day-list that shows available slots and greys out booked ones, a booking form to capture customer details, and a confirmation screen. Store bookings in localStorage so booked slots stay unavailable across refresh, and include a small admin view listing all bookings with a cancel action. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Booked slots must persist and become unavailable; double-booking is prevented in the UI. - Accessible date and time controls; responsive; clear confirmation with the booking summary. </constraints> <format> Return the full HTML as an artifact, then explain the availability and conflict logic and where calendar sync or payments would plug in. </format>
Builds a working booking-app MVP with real slot availability and an admin list as a previewable artifact.
Pro tip: Keep the slot length and hours small for the demo so the calendar fills up fast and the conflict logic is visible.
Service-Provider Listings MVP
9/30You are a marketplace engineer building provider-discovery apps (think rideshare-of-X or local pros). <context> I am testing a find-a-provider product. Build an MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, previewable as an artifact, where users browse providers and request one. </context> <inputs> - The service: [E.G. DOG WALKERS, TUTORS, MOVERS] - Provider fields: [NAME, RATING, PRICE, LOCATION, SPECIALTIES, BIO] - How users filter: [LOCATION, PRICE, RATING, SPECIALTY] - The request action: [BOOK, MESSAGE, REQUEST QUOTE] - Seed providers to preload: [5-8 EXAMPLES] </inputs> <task> Build a provider-discovery app: a filter bar, a sortable list of provider cards with rating and price, a provider detail panel, and a request flow that captures the user's details and saves the request to localStorage. Show a my-requests view of submitted requests. Preload the seed providers and persist all state. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; styled avatars as placeholders. - Filtering and sorting operate on live data; requests persist and show in the my-requests view. - Accessible cards and forms; responsive; clear empty states. </constraints> <format> Return the full HTML as an artifact, then describe the provider data shape and the matching or ranking logic you would build next. </format>
Builds a provider-discovery MVP with filtering, profiles, and a request flow as a previewable artifact.
Pro tip: Vary the seeded providers' ratings and prices so your sort and filter controls obviously change the results in a demo.
Rental / Equipment Listings MVP
10/30You are a product engineer who builds rental and peer-to-peer lending apps. <context> I want to validate a rentals idea. Build an MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, covering browse, availability, and a reservation flow. </context> <inputs> - What gets rented: [E.G. CAMERAS, TOOLS, PARTY GEAR] - Item fields: [NAME, DAILY PRICE, DEPOSIT, CONDITION, OWNER] - Rental period model: [BY DAY / BY HOUR] - Filters: [CATEGORY, PRICE, AVAILABILITY] - Seed items to preload: [5-8 EXAMPLES] </inputs> <task> Build a rentals app: a browse grid with filters, an item detail showing price, deposit, and condition, a date-range picker that computes the total cost, and a reserve action that stores the reservation in localStorage and marks the item as booked for those dates. Include a my-reservations view. Preload the seed items and persist everything. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; labeled color blocks for item photos. - Total cost is computed live from the date range and daily price; reservations persist and block overlapping dates. - Accessible date inputs and cards; responsive; clear reservation summary. </constraints> <format> Return the full HTML as an artifact, then explain the pricing and availability math and the deposit and insurance features a real launch would need. </format>
Builds a rentals MVP with date-range pricing and reservation tracking as a previewable artifact.
Pro tip: Test the math first — pick a multi-day range and confirm the live total matches daily price times days before trusting the rest.
Landing, Waitlist & Validation MVPs
5 promptsWaitlist + Landing MVP
11/30You are a growth engineer who builds pre-launch waitlist apps with viral mechanics. <context> I want to validate demand before building. Build a waitlist landing MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage so the signup and referral loop works with no backend, returned as a previewable artifact. </context> <inputs> - Product teaser: [WHAT IS COMING] - Who it is for: [AUDIENCE] - Reason to join early: [PERK] - Capture fields: [EMAIL, OR EMAIL PLUS NAME] - Referral mechanic: [E.G. MOVE UP THE LIST PER REFERRAL] </inputs> <task> Build a landing page with a benefit-led hero and an email-capture form that, on submit, stores the signup in localStorage and shows a success state with a position number, a referral link, and a progress meter toward a reward. Track a referral count in localStorage that moves the user up the list. Include a small how-it-works strip and a live signup counter derived from stored data. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - The form, position number, and referral counter genuinely work and persist across refresh. - Accessible form labels; responsive; the email field is the single dominant action. </constraints> <format> Return the full HTML as an artifact, then explain the referral logic and how to wire the email capture to a real provider before launch. </format>
Builds a waitlist landing MVP with a working referral loop and position counter as a previewable artifact.
Pro tip: Ask Claude to fake a believable starting count so the live counter reads like real momentum, not zero, on first visit.
Smoke-Test Landing With Fake-Door CTA
12/30You are a lean-startup engineer who builds smoke-test pages to measure intent before building. <context> I want to measure whether people would buy before I build anything. Build a smoke-test landing MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage that logs every click on a fake-door CTA, returned as a previewable artifact. </context> <inputs> - Product concept: [WHAT IT IS] - The promise and audience: [BENEFIT, WHO] - Pretend price or plan: [PRICE] - The fake-door CTA: [E.G. BUY NOW, START TRIAL] - What to ask after the click: [EMAIL, A ONE-QUESTION SURVEY] </inputs> <task> Build a convincing landing page with hero, value props, a pricing teaser, and a primary CTA that looks fully real. On click, instead of a real purchase, log the intent event to localStorage, then show a short capture step (email or a one-question survey) and an honest "we are gauging interest" message. Add a tiny hidden admin counter showing total clicks and captures so I can read the signal. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - The CTA, intent logging, and capture all work and persist in localStorage. - Accessible and responsive; the page reads as a real product, not a survey. </constraints> <format> Return the full HTML as an artifact, then explain what conversion signal to watch and how to be ethical about the fake-door test. </format>
Builds a smoke-test landing MVP that logs fake-door intent and captures interest as a previewable artifact.
Pro tip: Decide your pass threshold (e.g. 20% click-through) before running the test so the result is a real go/no-go, not a vibe.
AI-Wrapper UI MVP
13/30You are a product engineer who builds AI-wrapper apps — a focused UI around a single LLM use case. <context> I am building an AI tool that wraps a model for one job. Build the front-end MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with the API call clearly stubbed so I can demo the full UX. </context> <inputs> - The single job: [E.G. REWRITE EMAILS, GENERATE PRODUCT NAMES] - The input the user gives: [TEXT, FORM FIELDS, ETC] - Output format: [TEXT, LIST, CARDS] - Controls the user tweaks: [TONE, LENGTH, COUNT] - History behavior: [SAVE PAST RUNS YES/NO] </inputs> <task> Build the full wrapper UI: an input area with the tweakable controls, a generate button, a loading state, and a results panel that renders the output nicely with copy and regenerate actions. Stub the model call with a clearly-marked mock function that returns plausible sample output so the whole flow is demoable. Save run history to localStorage with a sidebar to revisit past runs. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Mark the mock API function with a comment showing exactly where to drop in the real fetch call and where the key goes (never hardcode a key). - History persists; UI is accessible, responsive, with copy-to-clipboard that works. </constraints> <format> Return the full HTML as an artifact, then explain the mock-to-real swap, the prompt you would send, and what to never expose client-side. </format>
Builds an AI-wrapper UI MVP with a stubbed model call, controls, and run history as a previewable artifact.
Pro tip: Have Claude make the mock output realistic for YOUR use case so testers react to the real experience, not lorem-ipsum results.
Concierge / Manual-Backend MVP
14/30You are a lean-startup engineer who builds concierge MVPs where a human fulfills behind a polished front-end. <context> I want to deliver the service manually but make the request experience feel like a real product. Build a concierge MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact. </context> <inputs> - The service delivered (by me, manually): [WHAT] - What the customer submits: [INTAKE FIELDS] - The stages a request moves through: [E.G. RECEIVED, IN PROGRESS, DELIVERED] - What the customer sees while waiting: [STATUS, ETA] - Audience and vibe: [WHO, AESTHETIC] </inputs> <task> Build a request app: a clean intake form, a confirmation with a request ID, and a status tracker the customer can revisit showing the current stage. Add a hidden operator view that lists all requests and lets me advance their stage — the change reflects in the customer status. Persist requests and stages in localStorage so the manual fulfillment loop is fully demoable. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Intake, status tracking, and the operator stage controls all work and persist. - Accessible forms and status UI; responsive; reassuring, specific copy. </constraints> <format> Return the full HTML as an artifact, then explain how the manual operator loop works and what to automate first once volume grows. </format>
Builds a concierge MVP with intake, status tracking, and an operator view as a previewable artifact.
Pro tip: Concierge MVPs win on trust — ask Claude to make the status tracker reassuring so manual delays feel intentional, not broken.
Mobile-App MVP Preview (Single Screen Flow)
15/30You are a mobile product engineer who prototypes app MVPs as phone-framed web previews. <context> I am validating a mobile app idea before building native. Build an interactive phone-framed MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, that runs the core flow inside a phone mockup. </context> <inputs> - App in one line: [WHAT IT DOES] - The core loop: [THE 2-3 SCREENS THAT MATTER] - Main object the user creates or tracks: [E.G. WORKOUT, EXPENSE, NOTE] - Bottom-nav tabs: [LIST 3-4] - Vibe: [AESTHETIC] </inputs> <task> Build a phone-framed app preview with a bottom tab bar and working navigation between screens. Implement the core loop for real: a list screen, a create flow, and a detail or summary screen, all persisting to localStorage. Include realistic mobile interactions (tap targets, a bottom-sheet create form, a simple stats tab). Make the core object fully creatable and viewable. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; centered phone frame with a max-width app viewport. - Tab navigation and the create-and-view loop genuinely work and persist. - Touch-friendly tap targets, accessible labels, smooth screen transitions. </constraints> <format> Return the full HTML as an artifact, then explain which native features the web preview can and cannot fake and what to build first in React Native or Swift. </format>
Builds a phone-framed mobile-app MVP preview with a working core loop as a previewable artifact.
Pro tip: Limit the preview to the two screens that prove the idea; testers judge a mobile MVP on the core loop, not the settings page.
Engagement & Niche App MVPs
5 promptsCommunity / Forum MVP
16/30You are a product engineer who builds community and discussion apps. <context> I want to test a niche community idea. Build a forum MVP as one self-contained HTML file with inline CSS, vanilla JS, and localStorage so posting and replying work with no backend, returned as a previewable artifact. </context> <inputs> - The community topic: [WHAT IT IS ABOUT] - Channels or categories: [LIST 3-5] - A post has: [TITLE, BODY, AUTHOR, TAGS] - Interactions: [UPVOTE, REPLY, BOOKMARK] - Seed posts to preload: [4-6 EXAMPLES] </inputs> <task> Build a forum with a channel sidebar, a feed of post cards (title, snippet, author, upvotes, reply count), a post-detail view with threaded replies, a new-post composer, and working upvote and reply actions — all persisting to localStorage. Add sort by new and top. Preload the seed posts so the community feels active on first open. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Posting, replying, and upvoting all persist and update counts live. - Accessible composer and feed; responsive; clear empty state per channel. </constraints> <format> Return the full HTML as an artifact, then describe the post and reply data shape and the moderation and notification features a real community would need. </format>
Builds a community-forum MVP with posting, replies, and upvotes as a previewable artifact.
Pro tip: Seed a couple of posts with replies already on them — an empty forum looks dead, a buzzing one invites the first real post.
Quiz / Assessment Tool MVP
17/30You are a product engineer who builds interactive quiz and assessment lead-gen tools. <context> I want a quiz that engages visitors and captures leads. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with scoring and a tailored result. </context> <inputs> - Quiz topic and promise: [E.G. WHAT IS YOUR FOUNDER ARCHETYPE] - Number of questions: [N] - The questions and options: [LIST OR ASK CLAUDE TO DRAFT] - Result types: [E.G. 3-4 OUTCOME PROFILES] - Capture model: [EMAIL BEFORE OR AFTER RESULT] </inputs> <task> Build a multi-step quiz: a start screen, one question at a time with a progress bar, scoring logic that maps answers to a result profile, an email-capture gate, and a tailored result screen with a shareable summary and a CTA. Save completions and captured emails to localStorage. Write the questions and result copy if I do not supply them. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Scoring, progress, and result mapping all work; captures persist in localStorage. - Accessible radio inputs and progress; responsive; honest, specific result copy. </constraints> <format> Return the full HTML as an artifact, then explain the scoring map and where to gate the email for the best completion-to-capture rate. </format>
Builds a quiz/assessment MVP with scoring, result profiles, and lead capture as a previewable artifact.
Pro tip: Gate the email right before the result, not before question one — people who finished the quiz convert far better.
Habit / Progress Tracker MVP
18/30You are a product engineer who builds tracking and streak apps. <context> I am validating a tracking-app idea. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with streaks and history that persist. </context> <inputs> - What users track: [E.G. HABITS, WORKOUTS, MOOD, READING] - Each entry records: [WHAT FIELDS] - Cadence: [DAILY / WEEKLY] - The motivating metric: [STREAK, TOTAL, AVERAGE] - Seed history to preload: [A FEW PAST ENTRIES] </inputs> <task> Build a tracker: an add-entry control, a list or calendar of logged entries, a streak and total summary, and a simple progress visualization (a CSS bar or dot grid built from the data). Editing and deleting entries works and recomputes the stats. Persist all entries to localStorage and preload the seed history so streaks are visible immediately. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Logging, streak math, and the visualization all derive from real persisted data. - Accessible controls; responsive; satisfying but honest feedback on logging. </constraints> <format> Return the full HTML as an artifact, then explain the streak calculation and the reminder and sync features you would add next. </format>
Builds a habit/progress tracker MVP with streaks, history, and a data-driven chart as a previewable artifact.
Pro tip: Preload a broken-then-rebuilt streak in the seed data so the streak logic is obviously working in your first screenshot.
Content Feed / Curation App MVP
19/30You are a product engineer who builds content-curation and feed apps. <context> I want to test a curated-feed product. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with save, filter, and a personal collection. </context> <inputs> - What the feed curates: [E.G. ARTICLES, TOOLS, DEALS, JOBS] - Each item has: [TITLE, SOURCE, TAGS, SUMMARY, LINK] - Filters and tags: [LIST] - Personal actions: [SAVE, HIDE, MARK READ] - Seed items to preload: [8-12 EXAMPLES] </inputs> <task> Build a feed app: a main feed of item cards, tag and source filters, a save action that adds items to a Saved collection, and a mark-read or hide action — all persisting to localStorage. Add a Saved tab showing the user's collection and a search box over the feed. Preload the seed items so the feed is full on first open. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Save, hide, mark-read, filter, and search all work on live persisted data. - Accessible cards and controls; responsive; clear empty state for the Saved tab. </constraints> <format> Return the full HTML as an artifact, then describe the item schema and how a real version would fetch and rank fresh content. </format>
Builds a content-feed MVP with saving, filtering, and a personal collection as a previewable artifact.
Pro tip: Mix sources and tags across the seed items so the filters visibly narrow the feed during a demo instead of returning everything.
Lightweight CRM MVP
20/30You are a product engineer who builds simple CRMs for solo founders and small teams. <context> I want to validate a niche CRM. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with a real pipeline and contact management. </context> <inputs> - Who the CRM is for: [E.G. FREELANCERS, REALTORS, COACHES] - A contact record has: [NAME, COMPANY, EMAIL, STATUS, VALUE, NOTES] - Pipeline stages: [E.G. LEAD, CONTACTED, WON, LOST] - The key view: [KANBAN BOARD / TABLE] - Seed contacts to preload: [4-6 EXAMPLES] </inputs> <task> Build a CRM with a pipeline view in the chosen layout where contacts move between stages (drag or a stage dropdown), an add-contact form, a contact detail with editable notes, and a summary bar showing count and total value per stage. Everything persists to localStorage. Preload the seed contacts so the pipeline looks real. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Moving a contact between stages, editing, and the value totals all work and persist. - Accessible board or table and forms; responsive; clear empty stages. </constraints> <format> Return the full HTML as an artifact, then explain the pipeline data shape and the integrations (email, calendar) you would build first. </format>
Builds a lightweight CRM MVP with a working pipeline and contact management as a previewable artifact.
Pro tip: Put a real dollar value on each seeded contact so the per-stage totals demo the one number founders actually care about.
MVP Planning Docs & Specs
5 promptsOne-Page MVP Spec
21/30You are a seasoned product lead who writes tight one-page MVP specs that engineers can build from. <context> I need a single-page spec that turns my idea into something buildable in two weeks. Produce it as a clean, formatted one-page document artifact (well-structured HTML or Markdown rendered as a doc) that I can hand to a developer. </context> <inputs> - Product idea: [ONE LINE] - The user and their core problem: [WHO, PAIN] - The one core user journey: [HAPPY PATH] - Must-have for v1: [WHAT CANNOT BE CUT] - Hard timeline or constraint: [E.G. 2 WEEKS, SOLO DEV] </inputs> <task> Write a one-page MVP spec with: a one-line product statement, the target user and problem, the single core user journey as numbered steps, an in-scope list and an explicit out-of-scope list, the data objects involved, a minimal success metric, and the riskiest assumption to test. Keep it to one page — ruthless and concrete, no fluff. </task> <constraints> - A single formatted document artifact that fits on one page; scannable headings. - Every line earns its place; the out-of-scope list must be as specific as the in-scope list. - Plain, decisive language — no buzzwords, no padding. </constraints> <format> Return the spec as a formatted artifact, then flag the single line in it most likely to be wrong and worth pressure-testing. </format>
Produces a ruthless one-page MVP spec with scope, journey, and risk as a previewable document artifact.
Pro tip: Force the out-of-scope list to be longer than the in-scope list; the cuts are what make a two-week MVP possible.
Feature-Cut List (MVP Scoping)
22/30You are a pragmatic product manager who specializes in cutting features to ship faster. <context> My MVP scope has crept and I need to cut. Produce a feature-cut decision document as a clean formatted artifact (a table-driven doc) that tells me exactly what to ship, defer, and kill. </context> <inputs> - Product and core promise: [WHAT] - The full feature wishlist: [PASTE THE LIST] - Hard constraint: [TIMELINE / BUDGET / TEAM SIZE] - The one outcome v1 must prove: [THE BET] </inputs> <task> Produce a MoSCoW-style cut list as a table: each feature rated Must / Should / Could / Won't for v1, with a one-line reason tied to the core bet and a rough effort tag (S/M/L). Below the table, give a tight "ship list" (the Musts only) and a one-paragraph rationale for the hardest cut. Be opinionated — default to cutting. </task> <constraints> - A single formatted document artifact with a clean, readable table. - Every rating has a reason that references the core bet, not personal preference. - No hedging — each feature gets a clear verdict. </constraints> <format> Return the cut list as a formatted artifact, then name the one feature you cut that will hurt the most and how to validate the idea without it. </format>
Produces an opinionated MoSCoW feature-cut list with a clear ship list as a previewable document artifact.
Pro tip: Paste your real, bloated wishlist verbatim — the value is in Claude defending each cut against your core bet.
MVP Build Plan (2-Week Roadmap)
23/30You are a technical lead who plans short, realistic MVP build sprints. <context> I have a scoped MVP and need a day-by-day build plan a solo founder or tiny team can follow. Produce it as a clean formatted document artifact with a timeline I can track against. </context> <inputs> - The MVP and its core loop: [WHAT IT DOES] - The stack or tools I will use: [E.G. NEXT.JS, SUPABASE, OR NO-CODE] - Team and hours per day: [WHO, CAPACITY] - Hard deadline: [DATE OR DAYS] - The single feature that proves the idea: [THE CORE] </inputs> <task> Write a day-by-day (or milestone) build plan for the MVP within the deadline: setup, the core loop first, then supporting features, then a polish-and-test buffer. For each block give the goal, the concrete deliverable, and a done-check. Front-load the riskiest part and explicitly park anything not needed to prove the core. End with a launch-day checklist. </task> <constraints> - A single formatted document artifact with a clear timeline and checkable deliverables. - The core loop is built before any nice-to-haves; include a real buffer for the unknown. - Realistic for the stated capacity — no superhero schedules. </constraints> <format> Return the build plan as a formatted artifact, then call out the one day most likely to slip and what to cut if it does. </format>
Produces a realistic day-by-day MVP build plan with done-checks and a launch checklist as a previewable document artifact.
Pro tip: Tell Claude your true daily hours; an honest 3-hours-a-day plan beats an optimistic full-time one you cannot keep.
MVP Validation Test Plan
24/30You are a lean-startup coach who designs cheap experiments to validate ideas before building. <context> Before I build, I want a plan to test whether anyone wants this. Produce a validation test plan as a clean formatted document artifact that lists the experiments, metrics, and go/no-go thresholds. </context> <inputs> - The idea and who it is for: [WHAT, WHO] - My riskiest assumption: [THE LEAP OF FAITH] - Channels I can reach users on: [E.G. REDDIT, EMAIL LIST, COLD DMS] - Time and budget I can spend: [CONSTRAINT] - What a real signal would look like to me: [INTUITION] </inputs> <task> Write a validation plan: restate the riskiest assumption as a falsifiable hypothesis, then list 3-5 cheap experiments (e.g. smoke-test page, concierge, fake-door, customer interviews) ordered by cost. For each, give the method, the metric, the go/no-go threshold, and the time to run it. End with a decision rule: what result greenlights the build and what kills it. </task> <constraints> - A single formatted document artifact; a clear experiments table plus the decision rule. - Every experiment has a numeric or binary threshold decided in advance — no vague "see if people like it". - Cheapest, fastest experiments first. </constraints> <format> Return the test plan as a formatted artifact, then name the single experiment to run this week and exactly what would make you walk away. </format>
Produces a validation test plan with falsifiable experiments and go/no-go thresholds as a previewable document artifact.
Pro tip: Make Claude set the kill threshold before you start; pre-committing to a number stops you rationalizing a weak result.
Lean Canvas (One-Page Business Model)
25/30You are a startup strategist who fills out crisp Lean Canvases for early ideas. <context> I want my whole MVP business model on one page. Produce a Lean Canvas as a clean formatted document artifact — the nine boxes laid out visually — so I can see the idea and its gaps at a glance. </context> <inputs> - Product idea: [ONE LINE] - Target customer segments: [WHO, EARLY ADOPTERS] - The top problems you solve: [1-3 PAINS] - How you make money: [REVENUE IDEA] - Your unfair advantage, if any: [WHAT] </inputs> <task> Fill in a Lean Canvas with all nine blocks: Problem, Customer Segments, Unique Value Proposition, Solution, Channels, Revenue Streams, Cost Structure, Key Metrics, and Unfair Advantage. Lay them out in the canonical grid. Keep each block to a few sharp bullets, and where I left something thin, make a clearly-marked assumption rather than leaving it blank. </task> <constraints> - A single formatted document artifact showing the nine-box grid visually. - Each block is concise and specific; flag assumptions with a marker so I know what to validate. - Honest — if the cost or revenue side is weak, show it rather than hiding it. </constraints> <format> Return the Lean Canvas as a formatted artifact, then name the two riskiest boxes and the order in which to test them. </format>
Produces a filled nine-box Lean Canvas with flagged assumptions as a previewable document artifact.
Pro tip: Ask Claude to mark every guessed box; the marked ones are your validation backlog, not facts to build on.
Most people use 10% of Claude. Tutorials unlock the rest.
AI Academy: 300+ hands-on tutorials on Claude, ChatGPT, Midjourney, and 50+ AI tools. New tutorials added every week.
Niche & Vertical MVPs
5 promptsCourse / Cohort Platform MVP
26/30You are a product engineer who builds learning and course-delivery apps. <context> I am testing an online-course product. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, where a learner progresses through lessons with saved completion. </context> <inputs> - Course topic and audience: [WHAT, WHO] - Modules and lessons: [LIST OR ASK CLAUDE TO DRAFT] - Lesson content type: [TEXT, VIDEO PLACEHOLDER, QUIZ] - Progress model: [MARK COMPLETE, PERCENT, CERTIFICATE] - Vibe: [AESTHETIC] </inputs> <task> Build a course player: a curriculum sidebar listing modules and lessons with completion checkmarks, a main lesson pane, mark-complete and next-lesson actions, and an overall progress bar derived from saved state. Lock lessons until prior ones are done if I want sequencing. Persist completion to localStorage so progress survives refresh, and show a finished state when all lessons are done. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; styled blocks as video placeholders. - Completion tracking, progress bar, and any sequencing all work and persist. - Accessible navigation and controls; responsive; clear current-lesson highlighting. </constraints> <format> Return the full HTML as an artifact, then explain the progress data shape and the auth, payments, and video hosting a real version needs. </format>
Builds a course-platform MVP with a lesson player and saved progress as a previewable artifact.
Pro tip: Turn on lesson sequencing in the demo so the locked-then-unlocked flow shows the product is more than a list of links.
Job-Board MVP
27/30You are a product engineer who builds niche job boards. <context> I want to validate a niche job board. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with browse, filter, post, and apply all working. </context> <inputs> - The niche: [E.G. REMOTE DESIGN, CLIMATE TECH] - A job posting has: [TITLE, COMPANY, LOCATION, SALARY, TYPE, TAGS, DESCRIPTION] - Filters: [LOCATION, TYPE, SALARY, TAGS] - Apply action: [APPLY FORM / EXTERNAL LINK] - Seed jobs to preload: [6-8 EXAMPLES] </inputs> <task> Build a job board: a search and filter bar, a list of job cards, a job-detail view, an apply flow that captures the candidate and stores the application in localStorage, and a Post a Job form that adds to the board live. Add a saved-jobs feature and a count of open roles. Preload the seed jobs so the board looks active. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; styled blocks as company logos. - Browse, filter, post, apply, and save all operate on live persisted data. - Accessible cards and forms; responsive; clear empty and no-results states. </constraints> <format> Return the full HTML as an artifact, then describe the posting schema and how you would handle employer accounts and paid listings. </format>
Builds a niche job-board MVP with browse, post, apply, and save flows as a previewable artifact.
Pro tip: Put real salary ranges on the seed jobs so the salary filter visibly works — it is the filter candidates use most.
Invoicing / Billing Tool MVP
28/30You are a product engineer who builds invoicing and billing apps for freelancers and small businesses. <context> I want to validate an invoicing tool. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, where creating and tracking invoices works end to end. </context> <inputs> - Who it is for: [E.G. FREELANCERS, AGENCIES] - An invoice has: [CLIENT, ITEMS, RATES, TAX, DUE DATE] - Statuses: [DRAFT, SENT, PAID, OVERDUE] - The summary I want: [OUTSTANDING TOTAL, PAID THIS MONTH] - Seed invoices to preload: [3-4 EXAMPLES] </inputs> <task> Build an invoicing app: an invoice list with status badges, a create-invoice form with dynamic line items that compute subtotal, tax, and total live, an invoice detail that looks like a real printable invoice, status changes (mark sent/paid), and a dashboard summary (outstanding, paid, overdue) derived from the data. Persist everything to localStorage and preload the seed invoices. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Line-item math, tax, totals, and the summary all compute correctly and update live. - The invoice detail is print-friendly; accessible forms; responsive; status badges are clear. </constraints> <format> Return the full HTML as an artifact, then explain the invoice and totals logic and where payment links and PDF export would plug in. </format>
Builds an invoicing-tool MVP with live line-item math, statuses, and a summary as a previewable artifact.
Pro tip: Stress-test the math by adding a line item with tax in the demo; if the total recomputes correctly, the core value is proven.
Budget / Expense Tracker MVP
29/30You are a product engineer who builds personal-finance and budgeting apps. <context> I want to test a budgeting product. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with categories, totals, and a simple chart all driven by real data. </context> <inputs> - Who it is for: [E.G. COUPLES, FREELANCERS, STUDENTS] - A transaction has: [AMOUNT, CATEGORY, DATE, NOTE, TYPE] - Categories and budgets: [LIST WITH MONTHLY LIMITS] - The key view: [LIST PLUS SUMMARY / BY CATEGORY] - Seed transactions to preload: [6-10 EXAMPLES] </inputs> <task> Build an expense tracker: an add-transaction form, a transaction list with category tags, a monthly summary showing spent vs. budget per category, an overall balance, and a simple CSS bar chart of spending by category built from the data. Editing and deleting recompute everything. Persist to localStorage and preload the seed transactions so the charts are populated. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only; chart built with CSS from real data. - All totals, budget comparisons, and the chart derive from live persisted transactions. - Accessible form and list; responsive; over-budget categories are visually flagged. </constraints> <format> Return the full HTML as an artifact, then explain the budgeting math and how you would import real bank transactions later. </format>
Builds a budgeting MVP with live category totals, budgets, and a data-driven chart as a previewable artifact.
Pro tip: Seed one over-budget category so the flagged-overspend state shows in your first demo and proves the app reacts to real numbers.
Inventory / Order Management MVP
30/30You are a product engineer who builds inventory and small-business operations apps. <context> I want to validate an inventory or order tool for small sellers. Build it as one self-contained HTML file with inline CSS, vanilla JS, and localStorage, returned as a previewable artifact, with stock levels and orders that update each other. </context> <inputs> - The business: [E.G. CRAFT SHOP, RESTAURANT SUPPLY] - A product has: [NAME, SKU, STOCK, PRICE, REORDER POINT] - Order fields: [PRODUCT, QTY, CUSTOMER, STATUS] - The alert I care about: [LOW STOCK, OUT OF STOCK] - Seed products and orders to preload: [A FEW EACH] </inputs> <task> Build an inventory app: a product table showing stock and a low-stock highlight, an add-product and restock control, an orders view where placing an order decrements stock, and a dashboard with totals and a low-stock alert list. Stock and orders stay in sync and persist to localStorage. Preload the seed data so the alerts and totals are meaningful on open. </task> <constraints> - One self-contained HTML file; no frameworks; Google Fonts only. - Placing or fulfilling an order updates stock live; low-stock alerts derive from reorder points. - Accessible tables and forms; responsive; cannot oversell below zero in the UI. </constraints> <format> Return the full HTML as an artifact, then explain the stock-and-order sync logic and the supplier and barcode features a real tool would add. </format>
Builds an inventory/order MVP with synced stock, orders, and low-stock alerts as a previewable artifact.
Pro tip: Set one product's stock just above its reorder point, then place an order in the demo to trigger the low-stock alert live.
Frequently Asked Questions
Prompts are the starting line. Tutorials are the finish.
A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.
7-day free trial. Cancel anytime.