01 · Content
Every asset has a page, an original description and a runnable demo
133 components, 74 prompts, 33 backgrounds — each with a detail page and a demo that mounts in place.
the catalog →Brand · launch
Launch checklists are usually advice. This one is a record: 12 items, each honestly labelled — 7 shipped, 3 left out on purpose, 2 still open. Every shipped item links to the page that proves it, so nothing here has to be taken on faith.
Shipped
7
Deliberately out
3
Still open
2
01 · Content
133 components, 74 prompts, 33 backgrounds — each with a detail page and a demo that mounts in place.
the catalog →02 · Content
The voice guide's banned list is the same dictionary the harness scans the public prose with.
voice guide →03 · Trust
Counters come from the data file, sizes from the build report, and gaps say 'not measured' instead of borrowing a score.
open metrics →04 · Trust
It runs on localStorage, feeds nothing, and says so on every panel that could be mistaken for live data.
05 · Access
Audited per asset; the catalog's banded scores are published rather than summarised.
audit register →06 · Access
A stylesheet-level rule with 0.001ms rather than animation:none, so end states are preserved.
craft notes →07 · Crawl
The export harness compares the sitemap's component count against the catalog at build time.
crawl surface →08 · Deliberately out
A demo signup would need fake data behind it. Until there is a reason and a backend, the button is the honest thing to omit.
09 · Deliberately out
Nothing here counts visitors, which is also why no page-view number appears on the metrics page.
why →10 · Deliberately out
The Pro pages describe what is built and mark the rest as not built; nothing takes a card.
11 · Open
No browser runtime was available to produce one; the speed page prints kilobytes and names the gap.
speed story →12 · Open
The changelog records size deltas per build; a weekly series would turn those into evidence rather than snapshots.
A checklist that only lists what shipped is a marketing page. The three deliberate omissions are decisions a reader can disagree with: no accounts (so no real favourites), no analytics (so no public page-view number), no paid tier (so no revenue). Each of those would have been easy to fake with a mock, and each would have made every other number on the site cheaper.
The two open items are the honest remainder: a Lighthouse run needs a browser this build never had, and a trend needs a second measurement. They stay on the list until they are done, not quietly dropped.
the env-tagged rows need a live domain; everything above them is verified offline