<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: riona he</title>
    <description>The latest articles on DEV Community by riona he (@riona_he_211b12cf8ae69e74).</description>
    <link>https://dev.to/riona_he_211b12cf8ae69e74</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4253162%2Fef7d1472-2258-4532-be42-a9a648353ea0.png</url>
      <title>DEV Community: riona he</title>
      <link>https://dev.to/riona_he_211b12cf8ae69e74</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/riona_he_211b12cf8ae69e74"/>
    <language>en</language>
    <item>
      <title>Stop Regenerating the Whole Page: Section-by-Section Landing Page Prompting in Cursor, Claude Code and Codex</title>
      <dc:creator>riona he</dc:creator>
      <pubDate>Fri, 20 Nov 2026 02:28:53 +0000</pubDate>
      <link>https://dev.to/riona_he_211b12cf8ae69e74/stop-regenerating-the-whole-page-section-by-section-landing-page-prompting-in-cursor-claude-code-3fb</link>
      <guid>https://dev.to/riona_he_211b12cf8ae69e74/stop-regenerating-the-whole-page-section-by-section-landing-page-prompting-in-cursor-claude-code-3fb</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure: I work on VibUI, a website prompt library. It gets one short mention near the end. The workflow works with prompts you write yourself.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Short answer:&lt;/strong&gt; when you add a landing page to an existing codebase with an AI agent, don't ask for the whole page in one go. Commit a small design-token file first, then prompt one section per request, each into its own component file, and review each diff before moving on. You get smaller diffs, fewer broken imports, and a page that matches the rest of your app. The pattern works the same way in Cursor, Claude Code, Windsurf, OpenAI Codex and GitHub Copilot agent mode.&lt;/p&gt;

&lt;h2&gt;
  
  
  What goes wrong when you ask an agent for the whole page?
&lt;/h2&gt;

&lt;p&gt;If you have ever typed "build a marketing homepage" into an agent inside a real repo, you have probably seen some of this:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;It installs a new animation or icon library you didn't ask for.&lt;/li&gt;
&lt;li&gt;It hardcodes twenty hex values that don't exist anywhere else in your app.&lt;/li&gt;
&lt;li&gt;It writes a 600-line &lt;code&gt;page.tsx&lt;/code&gt; that is painful to review.&lt;/li&gt;
&lt;li&gt;It "fixes" something unrelated in your layout while it's there.&lt;/li&gt;
&lt;li&gt;The second request ("make the pricing look better") rewrites the hero too.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;None of this is the model being bad. A single huge request gives it too much freedom and you too little control. Agents do their best work on narrow, checkable tasks, so give them narrow, checkable tasks.&lt;/p&gt;

&lt;h2&gt;
  
  
  What should I set up before the first prompt?
&lt;/h2&gt;

&lt;p&gt;Spend ten minutes on three things.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;1. A token file the agent must use.&lt;/strong&gt; For Tailwind v4 this can be a &lt;code&gt;@theme&lt;/code&gt; block; for older setups, &lt;code&gt;tailwind.config&lt;/code&gt; plus a few CSS variables. Keep it short:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight css"&gt;&lt;code&gt;&lt;span class="cm"&gt;/* app/marketing/tokens.css */&lt;/span&gt;
&lt;span class="k"&gt;@theme&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="nv"&gt;--color-ink&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mh"&gt;#111214&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nv"&gt;--color-paper&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mh"&gt;#f7f6f2&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nv"&gt;--color-accent&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mh"&gt;#2f6bff&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nv"&gt;--color-muted&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mh"&gt;#6b6f76&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nv"&gt;--radius-card&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;14px&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nv"&gt;--font-display&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"Instrument Serif"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;serif&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="nv"&gt;--font-body&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"Inter"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;sans-serif&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;strong&gt;2. A rules file.&lt;/strong&gt; Cursor reads &lt;code&gt;.cursor/rules&lt;/code&gt;, Claude Code reads &lt;code&gt;CLAUDE.md&lt;/code&gt;, Codex reads &lt;code&gt;AGENTS.md&lt;/code&gt;, Copilot reads &lt;code&gt;.github/copilot-instructions.md&lt;/code&gt;. Add a short "marketing pages" section to whichever you use:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Marketing pages live in app/(marketing)/. One component per section in components/marketing/.
Use only tokens from app/marketing/tokens.css. Never add new colours or fonts.
Do not install packages without asking. Do not edit files outside the paths named in the request.
Copy goes in [BRACKETS] unless I provide it. Never invent customers, logos or numbers.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;strong&gt;3. An empty page that imports placeholders.&lt;/strong&gt; Create &lt;code&gt;page.tsx&lt;/code&gt; with the section components imported but stubbed. Now every later request has an obvious place to land.&lt;/p&gt;

&lt;h2&gt;
  
  
  How do I prompt one section at a time?
&lt;/h2&gt;

&lt;p&gt;Each request names the file, the job of the section, its layout, and what it must not touch. Here's the hero:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Edit only components/marketing/Hero.tsx.
Purpose: explain in one line what the product does for a busy engineering manager.
Layout: two columns on desktop, stacked under 768px. Left: eyebrow label, headline
(max 9 words, font-display, 56px desktop / 38px mobile), one supporting sentence
(max 22 words, color-muted), primary button + text link. Right: a static screenshot
frame (rounded radius-card, 1px border ink/10%) with a [SCREENSHOT] placeholder.
Use tokens only. No new dependencies. No animation in this step.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Then the next one, in a fresh request:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Edit only components/marketing/Pricing.tsx.
Three plans in a row on desktop, stacked on mobile. Middle plan visually emphasised with
a 2px accent border, not a different background. Each card: name, price, one-line
description, 4 bullet points, button. Prices and bullets as [PLACEHOLDERS].
Add a monthly/yearly toggle as a controlled component; no external toggle library.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Notice what's missing: no adjectives like "modern", "sleek" or "premium". Those words give the model permission to guess. Measurements and constraints don't.&lt;/p&gt;

&lt;h2&gt;
  
  
  Which section order should I build in?
&lt;/h2&gt;

&lt;p&gt;Build in order of risk, not in page order:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Order&lt;/th&gt;
&lt;th&gt;Section&lt;/th&gt;
&lt;th&gt;Why this order&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;Hero&lt;/td&gt;
&lt;td&gt;Sets type scale and spacing that everything else copies&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;Pricing&lt;/td&gt;
&lt;td&gt;Most logic (toggle, plan data), so catch problems early&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;3&lt;/td&gt;
&lt;td&gt;Features&lt;/td&gt;
&lt;td&gt;Mostly layout, quick once the hero is right&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;4&lt;/td&gt;
&lt;td&gt;Social proof&lt;/td&gt;
&lt;td&gt;Placeholders only until you have real quotes&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;5&lt;/td&gt;
&lt;td&gt;FAQ&lt;/td&gt;
&lt;td&gt;Accessible accordion is a good isolated test&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;6&lt;/td&gt;
&lt;td&gt;Footer + nav&lt;/td&gt;
&lt;td&gt;Touches shared layout, so do it last and review carefully&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;7&lt;/td&gt;
&lt;td&gt;Motion pass&lt;/td&gt;
&lt;td&gt;One request that adds entrance motion to all sections at once&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The motion pass goes last on purpose. If motion is added section by section, each one tends to get its own easing and duration, and the page feels stitched together.&lt;/p&gt;

&lt;h2&gt;
  
  
  How do I review each diff quickly?
&lt;/h2&gt;

&lt;p&gt;A short checklist per section keeps reviews under five minutes:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Files touched:&lt;/strong&gt; only the one named? If not, reject and re-prompt.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;New imports:&lt;/strong&gt; anything from a package not already in &lt;code&gt;package.json&lt;/code&gt;?&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Raw values:&lt;/strong&gt; search the diff for &lt;code&gt;#&lt;/code&gt; and &lt;code&gt;px&lt;/code&gt;. Anything not from the token file is a bug.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Copy:&lt;/strong&gt; any invented names, logos, percentages or "trusted by 10,000 teams" lines?&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Mobile:&lt;/strong&gt; open devtools at 375px before accepting.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Keyboard:&lt;/strong&gt; tab through buttons and the FAQ. Focus rings visible?&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you use Claude Code or Codex in the terminal, ask for a summary of changed files at the end of each task. In Cursor and Windsurf, the diff view does this for you.&lt;/p&gt;

&lt;h2&gt;
  
  
  Does this work outside agent IDEs?
&lt;/h2&gt;

&lt;p&gt;Yes. The same "one section per request" habit works in browser builders:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Tool&lt;/th&gt;
&lt;th&gt;How to apply it&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Cursor, Windsurf&lt;/td&gt;
&lt;td&gt;Agent chat, one section per message, review the diff&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Claude Code, OpenAI Codex&lt;/td&gt;
&lt;td&gt;Terminal task per section; rules in CLAUDE.md / AGENTS.md&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;GitHub Copilot&lt;/td&gt;
&lt;td&gt;Agent mode in VS Code, rules in copilot-instructions.md&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;v0, Lovable, Bolt, Replit&lt;/td&gt;
&lt;td&gt;Ask for one section per message; pin tokens in the first message&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;ChatGPT, Claude, Gemini&lt;/td&gt;
&lt;td&gt;Generate a component per message, paste into your repo yourself&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Framer AI, Google Stitch, Readdy, Aura.build&lt;/td&gt;
&lt;td&gt;Generate a section, then adjust visually before exporting&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Where do section prompts come from if I don't want to write them?
&lt;/h2&gt;

&lt;p&gt;Writing the first few is the best way to learn the pattern. After that, it's mostly repetition: pricing tables, FAQs, feature grids and footers have the same handful of good layouts.&lt;/p&gt;

&lt;p&gt;That's the part VibUI covers. It's a library of 2,852 copy-paste website prompts, including 1,100 section prompts and 817 component prompts, plus 775 full-page prompts and 160 animated backgrounds. About 1,000 of them also include source code. Paid plans are one-time payments with no auto-renewal: $48 for a quarter, $99 for a year, or $179 lifetime. There are 60 free samples, and a free account can copy 2 prompts, at &lt;a href="https://vibui.dev" rel="noopener noreferrer"&gt;vibui.dev&lt;/a&gt;. If you'd rather keep everything in-house, the rules file plus a folder of your own section prompts gets you most of the way.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Should I let the agent create the token file too?&lt;/strong&gt;&lt;br&gt;
You can, but review it line by line. Every later section inherits its decisions, so it's the one file worth writing by hand.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What if the agent edits files I didn't name?&lt;/strong&gt;&lt;br&gt;
Undo, then repeat the request with "Edit only X. If another file needs changing, stop and tell me instead."&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Is one section per request slower?&lt;/strong&gt;&lt;br&gt;
The first draft takes a little longer. Fixing a 600-line generated page usually takes much longer than that.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Does this work with plain HTML instead of React?&lt;/strong&gt;&lt;br&gt;
Yes. Use one partial per section, or one &lt;code&gt;&amp;lt;section&amp;gt;&lt;/code&gt; per request with a comment marker, and keep the tokens as CSS variables.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Which model should I pick?&lt;/strong&gt;&lt;br&gt;
Any current coding model handles a single section well. Narrow requests matter more than which model you pick.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>ai</category>
      <category>nextjs</category>
      <category>productivity</category>
    </item>
    <item>
      <title>How to Prompt Any AI Builder for a Landing Page That Doesn't Look AI-Generated (Full Walkthrough)</title>
      <dc:creator>riona he</dc:creator>
      <pubDate>Thu, 19 Nov 2026 06:53:01 +0000</pubDate>
      <link>https://dev.to/riona_he_211b12cf8ae69e74/how-to-prompt-any-ai-builder-for-a-landing-page-that-doesnt-look-ai-generated-full-walkthrough-16dj</link>
      <guid>https://dev.to/riona_he_211b12cf8ae69e74/how-to-prompt-any-ai-builder-for-a-landing-page-that-doesnt-look-ai-generated-full-walkthrough-16dj</guid>
      <description>&lt;p&gt;&lt;strong&gt;Short answer:&lt;/strong&gt; write the prompt as a design brief in five blocks: &lt;strong&gt;Identity, Structure, Content rules, Motion, Never.&lt;/strong&gt; Put the restraint rules last, because the last constraints get followed most reliably. The same prompt then works in v0, Lovable, Bolt, Cursor, Windsurf, ChatGPT, Claude, Gemini, Replit or GitHub Copilot. Below is a complete example prompt you can copy, a line-by-line explanation, and how to paste it into each tool.&lt;/p&gt;




&lt;h2&gt;
  
  
  Why does the AI keep giving me the same landing page?
&lt;/h2&gt;

&lt;p&gt;A model writing a landing page has thousands of open decisions: font, weights, spacing, accent colour, section order, hero layout, motion. Each one you leave open defaults to the statistical middle. That's why so many AI landing pages share the same centred hero, gradient headline and three icon cards.&lt;/p&gt;

&lt;p&gt;You don't need a better model. You need to &lt;strong&gt;close the decisions&lt;/strong&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  What should a landing page prompt include?
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Block&lt;/th&gt;
&lt;th&gt;What it fixes&lt;/th&gt;
&lt;th&gt;Example line&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Identity&lt;/td&gt;
&lt;td&gt;colour, type, radius, borders&lt;/td&gt;
&lt;td&gt;&lt;code&gt;background #0A0A0B, one accent #FF5A1F, Inter 400/600 only&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Structure&lt;/td&gt;
&lt;td&gt;section order and layout&lt;/td&gt;
&lt;td&gt;&lt;code&gt;(1) Nav â€¦ (2) Hero â€¦ (3) Logos â€¦&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Content rules&lt;/td&gt;
&lt;td&gt;copy length and honesty&lt;/td&gt;
&lt;td&gt;&lt;code&gt;headline max 8 words; no invented testimonials&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Motion&lt;/td&gt;
&lt;td&gt;what moves and how&lt;/td&gt;
&lt;td&gt;&lt;code&gt;fade-up 16px, 500ms, stagger 60ms, respect reduced-motion&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Never&lt;/td&gt;
&lt;td&gt;the generic defaults&lt;/td&gt;
&lt;td&gt;&lt;code&gt;no gradient text, no emoji in headings, no carousel&lt;/code&gt;&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  The full example prompt (copy it)
&lt;/h2&gt;

&lt;p&gt;Here's a brief for a fictional developer tool called &lt;strong&gt;Tracebit&lt;/strong&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Build a single-page landing page for "Tracebit", a log search tool for backend developers.
Stack: React + Tailwind CSS, one file, responsive down to 360px.

IDENTITY
- Background #0A0A0B, text near-white at 92%, secondary text at 60%.
- One accent only: #FF5A1F. Used for the primary CTA, one highlight per section, focus rings.
- Font: Inter, weights 400 and 600 only. Monospace: JetBrains Mono for code.
- Radius 12px. Borders: 1px white at 8% opacity.

STRUCTURE
(1) Nav: wordmark left; 3 links + accent "Start free" button right; sticky; adds backdrop blur and bottom border after 24px scroll.
(2) Hero: left-aligned. Headline clamp(2.75rem, 6vw, 4.75rem), tracking -0.03em, max 8 words.
    Subline max-width 34rem at 60% opacity. Primary CTA + ghost secondary CTA.
    Right side: a terminal window showing 6 lines of a log query with one matched line highlighted in the accent.
(3) Logo row: 6 grey placeholder wordmarks at 40% opacity.
(4) Features: 3 rows alternating text/visual sides. Each visual is a bordered panel with a faint 24px grid background.
(5) One large testimonial quote in display type, with a placeholder name clearly marked [CUSTOMER NAME].
(6) Pricing: 3 tiers, middle tier elevated with accent border and "Most popular" tag.
(7) FAQ: accordion, one item open at a time, 5 questions.
(8) Final CTA band, then a minimal 2-row footer.

CONTENT RULES
- Use the placeholder copy in [BRACKETS] where real content is needed. Do not invent customer names, logos, numbers or reviews.

MOTION
- Sections fade up 16px over 500ms on first entry into view; children stagger 60ms.
- Cards lift 3px on hover. Respect prefers-reduced-motion (no movement).

NEVER
- No gradient text. No emoji in headings. No three-icon feature row. No stock illustrations.
- No carousel. No more than one accent colour. No centred hero.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Why does each line matter?
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;"One accent only" and "weights 400 and 600 only."&lt;/strong&gt; These are restraint rules. Left free, generators add a new colour and weight per section, and that's what reads as amateur.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;"Max 8 words" on the headline.&lt;/strong&gt; It forces real display-type hierarchy. Long headlines quietly shrink the hero.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A focal object in the hero (the terminal).&lt;/strong&gt; Heroes fail from emptiness more often than from clutter.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;"Do not invent customer namesâ€¦"&lt;/strong&gt; AI builders pad pages with fake social proof unless you say otherwise. Publishing that is a real liability.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;NEVER block last.&lt;/strong&gt; Trailing constraints tend to get the highest obedience, so the anti-generic rules go at the end.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How do I use this prompt in each AI tool?
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Tool&lt;/th&gt;
&lt;th&gt;Where to paste&lt;/th&gt;
&lt;th&gt;Tip&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;v0&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;New chat&lt;/td&gt;
&lt;td&gt;Ask for "one file, no extra libraries" to keep it portable&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Lovable&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;First message of a new project&lt;/td&gt;
&lt;td&gt;Keep design and features in separate messages&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Bolt.new&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Main prompt box&lt;/td&gt;
&lt;td&gt;Ask it to put colours in the Tailwind theme first&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Cursor / Windsurf&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Agent chat with your repo open&lt;/td&gt;
&lt;td&gt;Name the target file, e.g. &lt;code&gt;app/(marketing)/page.tsx&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Claude / Claude Code&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Chat (Artifacts preview) or terminal agent&lt;/td&gt;
&lt;td&gt;Ask for a single React component first, then split it up&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;ChatGPT&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Chat, then open in Canvas&lt;/td&gt;
&lt;td&gt;Iterate on copy in Canvas without regenerating the layout&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Gemini / Google AI Studio&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Chat or Build mode&lt;/td&gt;
&lt;td&gt;Build mode gives you a live preview and a shareable link&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Replit&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Replit Agent&lt;/td&gt;
&lt;td&gt;It can host the result immediately&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;GitHub Copilot&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Copilot Chat / agent mode in VS Code&lt;/td&gt;
&lt;td&gt;Paste into a new branch so you can diff the result&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  How do I iterate without breaking the design?
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Swap content, not structure.&lt;/strong&gt; Replace the [BRACKETS] with real copy first.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Change one decision per message.&lt;/strong&gt; "Make the accent #3B82F6" is safe. "Make it more modern" undoes your brief.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Check 360px width, keyboard focus and dependencies&lt;/strong&gt; before you accept it.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Where can I get more prompts like this?
&lt;/h2&gt;

&lt;p&gt;Writing a brief like this takes 30â€“60 minutes once you know the pattern, and much longer the first few times. If you'd rather start from a tested one, &lt;a href="https://vibui.dev" rel="noopener noreferrer"&gt;VibUI&lt;/a&gt; has &lt;strong&gt;2,800+ website prompt templates&lt;/strong&gt; written in this style: 775 full pages, 1,100 sections, 817 UI prompts and 160 animated backgrounds, each with a live preview. About 1,000 also ship with finished source code.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;It's plain text, so it works in every tool in the table above&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;One-time payment, no subscription:&lt;/strong&gt; $48 (90 days), $99 (1 year, includes source code), $179 (lifetime, listed down from $384)&lt;/li&gt;
&lt;li&gt;60 free samples you can browse without an account at &lt;a href="https://vibui.dev/free" rel="noopener noreferrer"&gt;vibui.dev/free&lt;/a&gt;, and a free account can copy 2&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Does this work with ChatGPT and Claude, or only v0?&lt;/strong&gt; All of them. The prompt is plain text. Any tool that writes front-end code from a prompt can use it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Should I specify the framework?&lt;/strong&gt; Yes, if you have one. Otherwise say "React + Tailwind, one file" for portability, or "plain HTML + CSS" for ChatGPT or Claude previews.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Why not just use a component library?&lt;/strong&gt; Components give you parts. The brief gives the AI the layout, type, colour and motion decisions that make the parts look designed. They work well together.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How long should a landing page prompt be?&lt;/strong&gt; Usually 250â€“500 words. Length isn't the goal. Closing open decisions is.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: written by the VibUI team. The Tracebit prompt above is an original example written for this tutorial.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>webdev</category>
      <category>tutorial</category>
      <category>beginners</category>
    </item>
  </channel>
</rss>
