Framer

Promptfree, for everyone, and the only version there is

Copy the prompt for Claude CodeCopy the prompt for CodexCopy the prompt for Cursor

Copy

Build me the website I actually need instead of Framer — and know what the verdict means.

What you are paying for is **a visual canvas whose output is genuinely good markup**, plus hosting and a CMS in one subscription. Reproducing the canvas is a project of its own. The honest alternative is not another builder: it is a static site generator with a small content layer, which produces better markup than any canvas and costs nothing to host. That trade — losing the canvas, keeping the quality — is why this is ALMOST.

STACK
- Astro, static output where possible and a Node adapter where a form needs a server
- SQLite through better-sqlite3 only if there is a content layer or form handling
- Vanilla CSS and a little vanilla JS. No client-side framework
- Caddy in front, with automatic TLS

THE SITE
- Sections as components, each taking its content from data rather than having it embedded: hero, features, gallery, testimonials, pricing, FAQ, contact, footer
- Content in Markdown files with front matter, or in a small SQLite-backed editor if somebody non-technical maintains it. That editor is what replaces the CMS, and it is one screen per content type
- Under 100KB for a page including the font, rendering complete in a second on a phone over a poor connection. A canvas builder rarely achieves that, which is the quiet advantage of doing it this way

TYPE AND COLOUR
- One self-hosted variable font as WOFF2, subset to the characters used. Never a font from a third-party CDN
- A small token set — two backgrounds, two text colours, one accent, one border — with every pair measured against WCAG AA rather than chosen by eye
- One size scale and one spacing scale, used everywhere. That constraint is what makes a hand-built site look designed

THE THINGS A BUILDER GIVES YOU FOR FREE, WHICH YOU MUST NOT SKIP
- Responsive from 320px with no horizontal scrolling and nothing hidden to make it fit
- Open Graph and Twitter tags, with a share image generated at build time from the page's own title using satori and resvg
- A favicon set from one source image, at the sizes that are actually used
- JSON-LD for the page type — Organization, Product, Article, FAQPage
- Sitemap, robots.txt, canonical URLs, and a 301 for every changed path
- Semantic HTML: one h1, real landmarks, labels tied to inputs, visible focus rings, `prefers-reduced-motion` respected
- Images with dimensions set, served as WebP or AVIF with a fallback, lazy below the fold, and resized to the sizes actually used
- A link checker in CI, so a dead link is a failing build

THE FORM
- Fields defined in data, validated on the server, stored in SQLite and emailed on
- A honeypot and a minimum time on form, plus a rate limit per address hash. No third-party form service and no captcha that sends your visitors elsewhere
- Works without JavaScript: a plain POST, a redirect, a message

ANIMATION, WHICH IS WHY PEOPLE CHOOSE THE CANVAS
- Scroll-driven animations and view transitions are now CSS features. Use them rather than a library
- Respect `prefers-reduced-motion` on every one, completely
- Restrain it. The reason a canvas-built site often feels heavy is that everything moves

DEPLOYING
- Built in CI on every merge, deployed as static files
- Cache headers: hashed assets immutable, HTML short
- A content security policy strict enough to forbid third-party scripts, which is easy here because there are none
- A one-line deploy script and the domain setup written out

WHAT MATTERS MOST
Speed and the search-engine surface. Get the page under 100KB and every meta tag right, then compare it against the builder's output on a phone. If the hand-built one is not faster and cleaner, there was no reason to leave.