Examples
Eight sites, built from nothing but <include>,
layouts, slots, the underscore, and the .fragment.html opt-out. Four
were built by an agent that had never seen unify before, given only
the
sixty-line authoring rules and a client brief. Five build with no dependencies
at all and run live below; two need an npm toolchain and link to their source for
now; the eighth is this site.
Live in this deployment
Dependency-free: each one builds with nothing but unify build, so
the deploy workflow renders a real copy alongside this site.
seed-library
Fernhollow Seed Library: a community seed-library site — 41 pages built from a 27-variety catalogue, three growing guides, and a seasonal-notes section with its own layout. The catalogue page renders every variety at build time and filters client-side with a small inline script, so there is no fetch to 404 on a subpath deploy.
seed-library-alt
The same Fernhollow Seed Library brief, solved by a second, independent
author: 42 pages from the same 27-variety catalogue, composed with a different
mix of layouts and includes. Built side by side with seed-library
to see how differently two authors can compose the same five primitives from
one brief — and, deliberately kept, one privacy mistake the other two avoid:
it publishes the catalogue's private seed-keeper field on every variety page.
See "What
these examples do not show" in the repo README.
seed-library-ondemand
A later, larger Fernhollow brief at real scale: 225 varieties across 12 families, 232 pages. Rather than shipping the whole catalogue into the page, the browse UI fetches one family's JSON at a time through real anchors carrying the fetch URL, so the deploy address never has to be hardcoded in JavaScript.
htmx-fragments
Hand-maintained to document the .fragment.html opt-out working
with htmx: month lists are bare fragments a button swaps in with
hx-get, and an opening-hours panel is one file consumed twice —
spliced into the Visit page by <include> at build time, and
fetched raw by htmx at runtime. htmx itself is vendored, so nothing loads from
another origin.
catalog-search-blog
The runnable companion to the catalog-and-search guide: Fieldnotes, a
five-post blog whose entire listing page — post list, tag facet, series facet,
and search box — renders client-side from catalog.json and
search-corpus.json, the two files --catalog/
--search-corpus write once at build time. No framework, no server,
no dependency at all.
Also in this repository
Two more examples need an npm toolchain to build, so they are not part of this deployment yet — that is a follow-up after 0.9.0 publishes. And the eighth is the site you are reading now.
eleventy-htmx
Ashgrove Instruments: a product front page, documentation pages, and a
release-notes stream filtered by topic — rendered by Eleventy running inside
unify's --generate seam, then composed into unify's layout,
reference-checked and published exactly like a hand-written page. Needs
npm install && npm run build, so it isn't live here yet.
forge-svelte
Thistleknap Forge: a fee-estimator widget authored as a Svelte component,
compiled by a small _scripts/ build into one plain JS bundle that
unify ships as an ordinary asset — unify needs to know nothing about Svelte.
Needs an npm toolchain to build, so it isn't live here yet.
unify-docs
unify's own documentation site — you are looking at it. Built by unify from
the repository's real docs/ directory through
--generate, so it cannot drift from the docs it renders, and it
exercises the full production layer: pretty URLs, a base URL, completed
canonicals, a sitemap, a catalog, and a search corpus.