How to use the section prompt builder
- Pick a section type, for example Pricing, Cart drawer or Hero. The rest of the form adapts to it.
- Choose a layout variant and the content slots the section needs, such as headline, CTAs and social proof for a hero.
- Fill in the specific details that appear for some sections: pricing tiers, hero copy, form fields or the free-shipping threshold for a cart.
- Review states and interactions. States are the moments the design must cover, such as empty, loading, error and success. Built-in best practices are pre-selected.
- Set the stack, style, palette, font and output format, then choose the target AI and detail level.
- Copy, download or share the prompt. Save combinations you reuse as presets.
How it works: what a section prompt needs
A section is small, but the details that make it production-ready are easy to leave out. Every prompt this builder generates covers:
- Role and task: a stack-specific role (“a senior React + Tailwind developer”) and a one-line task naming the section and where it will be used.
- Layout and content: the variant, the content slots, the number of items where relevant, breakpoint behaviour and where the copy comes from.
- Section-specific details: tiers with prices, period and a highlighted plan for pricing; headline and CTA text for a hero; field lists and spam protection for forms; a free-shipping bar and upsell for a cart.
- Interactions and states: the states to design, the interactions to support and the accessibility rules for that pattern, following the W3C’s ARIA Authoring Practices where they apply.
- Design and output format: style, palette roles, radius and font, plus exactly what to return — a typed component, HTML and CSS, Tailwind JSX or a Shopify section with schema — and an optional usage example.
- Acceptance criteria at detail level 3 and above: no horizontal scroll at 360px, zero critical axe issues and no layout shift when states change.
Worked examples
Example 1: a SaaS pricing table for ChatGPT
Apply SaaS pricing, 3 tiers and choose ChatGPT at detail 3. The prompt asks for a responsive pricing component with Starter ($12/mo), Pro ($29/mo, “Most popular” badge and highlighted border) and Business ($79/mo), six feature rows with screen-reader text for each tick, and a monthly/yearly toggle showing “Save 20%”. It requires the toggle to be a keyboard-operable radio group that announces price changes with aria-live, and ends with a scope block.
Example 2: a Shopify cart drawer for Claude
Apply Shopify fashion cart drawer with the Claude target. The prompt wraps each part in XML tags and asks for a .liquid section with line items, quantity steppers, an upsell for a matching scarf and a free-shipping progress bar towards $100. Its built-in rules require a focus trap that closes on Esc and returns focus to the cart button, and the output format adds a schema block with settings, blocks, presets and translation keys.
Example 3: a hero section for v0
Apply SaaS hero with product mockup and switch the target to v0. The prompt lists the eyebrow, the headline “Answer every customer in minutes”, two CTAs, social proof and a product screenshot. It includes the rule that the hero image is the Largest Contentful Paint element and must load eagerly with fetchpriority=“high”, which protects page speed scores. Because v0 builds visual components, the prompt ends with an invitation to ask questions before building.
Tips and common mistakes
- Name the context. “For a Shopify fashion store” changes the copy, the imagery and the conventions the AI follows.
- Always cover states. Empty carts, failed submissions and loading tables are where generated components most often break.
- Keep the built-in rules. They encode accessibility requirements that are expensive to add later.
- Match the output format to your stack. Ask for a Liquid section for Shopify and typed props for React, so the code drops into your project without rework.
- Reuse your palette. Use the same palette and radius as your Website Prompt Builder output so every section matches.
Limitations
The builder produces prompts only; it doesn’t generate or test code. AI tools sometimes ignore parts of a prompt, so check each acceptance criterion yourself, especially keyboard operation and screen-reader announcements. Built-in rules cover common patterns but are not a substitute for an accessibility audit of the finished site.