The landing page for ux-ui-agent-skills, built with the kit it is advertising.
Static HTML, no framework, no build step for development. The scroll choreography comes from a portfolio template; everything on top of it, the proof strip, the design-system wall, the how-to-use section and the install block, is new and driven by the kit's own design tokens.
python3 tools/serve.py 3000 # sends no-store, so edits show on reload
Then open http://localhost:3000.
The project was created as a Worker, not a Pages project, so it deploys with
wrangler reading wrangler.jsonc. In the dashboard:
| Setting | Value |
|---|---|
| Build command | bash tools/build.sh |
| Deploy command | npx wrangler deploy |
| Root directory | / |
The build has to run first, because wrangler.jsonc serves ./dist and
nothing else. With the build command left as None the deploy fails on a
missing directory.
Custom domain: add it under the Worker's Domains tab, then point the site at it in one command.
python3 tools/set_domain.py skills.designlazyyy.com
That rewrites the canonical URL, the sitemap, the OG tags and robots.txt
together, so they cannot drift apart.
tools/build.sh copies only the web files into dist/. The kit itself
(tokens/, components/, scripts/, CLAUDE.md and the rest) stays in the
repo, where it is the point, but never reaches the CDN.
_headers sets the cache policy and the security headers, _redirects folds
/index.html and /about onto their canonical URLs, and 404.html is served
for anything else (not_found_handling: "404-page"). All three are copied into
dist/ by the build.
The canonical URL, the sitemap and the OG tags are all set to
https://skills.designlazyyy.com. Change that one string inrobots.txt,sitemap.xml,index.htmlandabout.htmlif a custom domain is attached.
The page is not just about the kit, it is gated by it. Every value in
public/site.css resolves through public/theme.css, which is generated from
tokens/*.json.
python3 tools/build_theme.py # tokens/*.json -> public/theme.css
python3 scripts/lint_hardcodes.py public/site.css # no hex, px or timing off the theme
python3 scripts/validate_theme_refs.py public/theme.css public/site.css
python3 scripts/check_no_emoji.py index.html about.html public/site.css
python3 scripts/validate_contrast.py # WCAG 2.2, light and dark
node scripts/verify_responsive.mjs index.html # no sideways scroll at 280/320/414
Two real bugs came out of running these against the page: --space-2.5 is not
a legal custom-property name, so every half-step token was failing silently,
and a <pre> in a grid item without min-width: 0 was scrolling the whole
document sideways on a phone.
images/design-systems/ holds 61 homepage screenshots, captured locally with
headless Chrome. They are the live web, not mockups.
./tools/capture.sh 0 1 # capture everything in tools/design-systems.tsv
python3 tools/build_wall.py # regenerate the wall markup in index.html
python3 tools/build_meta.py # regenerate the social card and the favicon set
77 of the library's 138 entries have no screenshot: most are aesthetic archetypes with no address, and a handful of hosts answer a headless browser with a consent wall. Those are listed by name instead of shown.
MIT. The screenshots are of third-party websites and remain the property of their owners; they are used here to illustrate the design systems the kit specs.