Every page on ten emulated phones and tablets: 190 loads, 21 blocking and 530
high findings, on pages that had passed Batch 11 and Batch 16.
#214 was fixed in v0.9.5 and released still broken. The check was a desktop
window sized to 768, which has a scrollbar, so the layout viewport was ~753px
and the md breakpoint the fix was about never engaged. At a true 768 the brand,
six nav links and the CTA want 787px against 736px of container, so no amount of
gap tightening closes it. The desktop row now starts at lg, and 768 to 1023 gets
the Sheet menu, which is the better tablet experience regardless: 44px rows
instead of 17px ones, and submenus that do not depend on hover. The CTA takes
shrink-0 and whitespace-nowrap, so the next row that does not fit overflows
visibly instead of being squeezed past the edge while the total still measures
as fitting.
The Cisco mark fills 66% of its 700x700 canvas, so both tiles scaled it up
inside overflow-hidden and cut 13px off the trademark on / and 24px on /about.
Cropping the viewBox to the artwork renders it at 62x46 on /, the same size as
the 8x8 logo beside it, unscaled and unclipped.
491 standalone links were 17 to 26px tall. .tap-target grows the hit box and
takes it back out of the layout, and every list using it moves to space-y-4:
at space-y-2 neighbouring boxes overlap and measure 33px while not being there.
Footer links also take block, so the row is the target. A 39x36 word is judged
as a compact target wanting 44px; a 200x36 row wants 32.
scripts/device-sweep.mjs and scripts/lib/css-audit.js are the instrument. They
stay out of verify.sh: playwright is global here and the sweep needs a running
server, and a guard that cannot run on a clean clone is one that gets skipped.
After the fixes: zero blocking, zero high across all ten devices. Each defect
was then re-introduced and reported again (clipped x1, media_overflow x1,
past_viewport x19, touch_target x342).
closes#238closes#239closes#240closes#241
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Three things, all in the path between this repository and the running image.
Closes#225, #224 and #223.
1. THE PRODUCTION SITEMAP CARRIED NO LASTMOD AT ALL. Dates come from git
history, and the image build cannot see git: .dockerignore excludes .git and
node:alpine has no git binary. prerender.js read the failure into an empty
catch commented "git unavailable or file untracked", so all 18 URLs came out
undated while the build printed a success line. Local builds looked perfect,
which is why nobody caught it.
release.sh now computes the map where git exists, passes it as the
SITEMAP_LASTMOD build arg, and then asks the built image whether its sitemap
has dates, refusing to publish one that does not. prerender prints the count
on every run, so "18 URLs, 0 dated" can never again read as success. The
route-to-source map moved into scripts/lib/routes.js, where a service page
now also counts its own content file, so editing one page's copy moves that
page's date and no other.
Proven: an image built with the arg carries 18 lastmod entries; a build with
git deliberately unreadable and no arg reports "18 URLs, 0 carrying a
lastmod" and warns.
2. THE DOCKER BUILD CONTEXT CARRIED CLIENT MATERIAL AND LIVE SECRETS. .drop/,
zoho.md (the reCAPTCHA secret and the Zoho tokens), Levi.md and two 30 MB
zips were all sent to the daemon on every build, along with four agent
workspaces. The final image copies only built output, so none of it ever
shipped, but one careless COPY would have changed that. Proven by listing the
context from inside a throwaway image: before, all of it; after, none of it.
3. UNTRACKED FILES PASSED SILENTLY. docker build packs the working tree, so an
untracked module the code imports produces an image that works and a tag that
cannot rebuild it. release.sh now refuses while untracked files are present,
and pre-commit's note counts them too.
#223 also claimed post-commit hides a refused push. It does not: it printed
"push was refused. The commit is safe locally and the branch is now ahead."
during this batch. The issue was corrected on the tracker rather than acted on.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Everything here checked an input: the content check reads the data, the secret
scan reads the diff, the build reads the source. Nothing read the OUTPUT, which
is the only thing a visitor or a crawler ever sees. Two live defects made the
case: every page preloaded the wrong image for months, and eleven pages shipped
a run-on description. Both are plain in the built HTML and invisible in the
source.
Build mode is guard 15-built-html, after 10-build. Per page it requires exactly
one title, one non-empty description, one canonical equal to the site origin plus
the route, and one h1; JSON-LD that parses, with no FAQPage, which the owner
ruled out; no em dash and no U+FFFD; a preload naming the image the page actually
paints first; and no description that runs its short description into the next
sentence. Across pages it requires every internal link and every fragment to
resolve, the sitemap to list exactly the routes the site serves, and 404.html to
carry noindex and no canonical. It exits 2, not 0, when dist/ is missing or older
than the sources: auditing stale output is auditing nothing.
That also guards a specific hazard. react-helmet-async on React 19 does not
merge, so a second <SEO> anywhere on a page silently emits a second title and a
second canonical, and a search engine picks whichever it likes.
URL mode fetches every page in a live sitemap once per crawler user agent
(OAI-SearchBot, PerplexityBot, ClaudeBot, Googlebot, bingbot), requires HTTP 200
and identical bytes across agents, runs the same page rules, and reports any URL
without a lastmod. It is deliberately NOT wired into deploy.sh: a check that runs
after publication cannot stop it, and pretending otherwise is worse than not
having it. Run it after a deploy.
Proven by mutation, nine of them, each restored afterwards: a wrong canonical
(7 pages), a second h1 (4), FAQPage markup, an em dash in copy, a link to a route
that does not exist (18), a fragment that is not on its target page (7), the
preload keyed on the old attribute (19), the template title left in place giving
two titles (19), and the industry routes dropped from the route list (56). A
clean build audits clean, and URL mode passes against the local server as all
five crawlers.
Closes#228.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
src/data is prose in a data structure, and nothing checked it. The long-form
service pages make that dangerous in a specific way: their copy arrives as an
owner-approved markdown sheet that MIXES DIRECTIONS TO THE WEBSITE MANAGER INTO
THE COPY. "Do not promise that every number is always portable." "Keep this
factual:" "Place an official 8x8 Work screenshot beside this section." Those
lines look exactly like copy, and publishing one puts an internal instruction on
a customer-facing page.
scripts/lib/content.js decides whether the content layer is publishable, and
prerender.js runs it before rendering anything, so every build enforces it: the
pre-commit hook, npm run verify, and the Docker image build. It refuses a
website-manager direction, an em dash, a U+FFFD, markdown or an HTML tag left in
a string, an unknown block type, a section id that is not letter-first, unique
and free of the layout's own ids, a section that does not open with its direct
answer (unless it declares kind list or faq), a FAQ question with no answer, a
link to a route or fragment that does not exist, an image whose src is missing
from public/ or has no alt or no dimensions, and the missing benefits or
idealFor list that the short layout maps without checking. A description over
160 characters is a note, not a failure: owner-approved copy is published as
written.
scripts/lib/routes.js is now the one route list. prerender.js built its own
while src/routes.jsx built the router's, and nothing compared them: a route in
one and not the other is never prerendered, so the server answers it with
404.html while the site's own navigation links to it. entry-server.jsx exports
the router table so the build can compare the two.
Proven by mutation, seventeen of them, each expecting exactly one finding and
getting it: unknown block type, FAQ answer removed, answer moved below its list,
duplicate id, digit-leading id, link to /services/contact-centre,
#no-such-section, missing image file, image without dimensions, image without
alt, em dash, a manager direction, markdown bold, U+FFFD, an HTML tag, missing
h1, empty section. Both generated content modules pass unmutated. Against a real
build: an em dash added to industries.js failed npm run build naming the field,
and a /pricing route added to src/routes.jsx failed it naming the route.
Closes#230.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
null