website: the operator's seven notes - a services page, the hero, and 100 dead links

Reported from his own screens. Five defects, two design.

1. THE CATEGORY ANCHORS WERE DEAD - all 100 of them. The sidebar's five app-category
   links pointed at #fajlkezeles-irodai and friends; the ids are #fajlkezeles and
   friends. The sidebar is copied into every page, so one wrong fragment was wrong
   twenty times, and a dead fragment raises no error anywhere: the browser stays put.

   New gate 23 resolves every in-site #fragment against the page it names (360 today).
   It convicted two more on its first run: the home page's four hero tiles pointed at
   /alkalmazasok#immich, which no page had ever carried, and the FAQ's contact button
   pointed at a home-page section that had moved out. Every app card now carries its
   slug as an id - taken from the card's own logo file, which gate 21 already ties to
   the catalog, so the ids cannot drift from it silently.
   Decoy site/dead-fragment puts the operator's own bug back and the gate convicts.

2. SZOLGALTATASOK IS A PAGE: /szolgaltatasok + /en/services, registered in TWINS, URL
   and SHELL. The ten capability items moved VERBATIM out of the home page; the home
   keeps the promise sentence, the three group names as links and a button, and is 345
   Hungarian / 434 English words shorter. The .caps rules stopped being .page-index-
   scoped, because the block now lives on a page with a different body class.

3. THE HOME HEADLINE is back in the hero column beside the picture, two lines with the
   second in --blue-bright at clamp(2.1rem, 4.4vw, 3.2rem). M3 had moved it into
   .page-header - the INNER pages' heading - so it shrank to 2.1rem and the domain
   badge was stranded beside it.

4. THE HERO PICTURE is the real dashboard in a laptop. The frame is the operator's own
   supplied render; the screen is dashboard-start-{hu,en}.webp composited into it, so a
   visitor sees a real box's Launcher rather than a drawing of one - the supplied
   screens are AI-drawn and misspell app names. 57 KB, named dashboard-laptop-<lang>
   so gate 19 holds it to the page's language.

5. .page-lede HAD NO CSS AT ALL on 16 pages: a bare paragraph, full width of the
   content column, hard against the box beneath it. 760px measure, centred over the
   box, air beneath; left-aligned on /szolgaltatasok, which has no centred box.

6. THE SIDEBAR LOGO: the wordmark shipped at 0.67x the mark's width where the real
   lockup (assets/logo.png, measured: mark 492px wide, wordmark 631px) has it at 1.28x.
   Mark 112px, wordmark 25px tall, 28px less vertical padding.

7. SCROLLBARS are the page's own now. The first attempt missed the viewport's: it is
   painted from <html>, and a `body.shell ::-webkit-...` rule only reaches scrollers
   inside the page.

Also: the two Latin font subsets are preloaded. With font-display: swap the page
reflowed after first paint - measured at 1339px on /alkalmazasok#media.

Checked at 390px and 1440px on all 21 pages: no sideways scrolling, footer inside the
content column everywhere, drawer opens with all 13 items. site.css?v=24.

scripts/test_gate_decoys.py

@@ -91,7 +91,7 @@ COVERS = {
                         "switch at the wrong twin, lang=\"hu\" on an English page, an app missing from one apps page, "
                         "ASCII-only Hungarian in English text and in a script message, a 'please', an English "
                         "retrieval promise the Hungarian does not make, a FAQ JSON-LD drifting from the visible "
-                        "answer, text after </html>, a Hungarian dashboard picture on an English page, and four picture-viewer faults (an opener without href, an opener to a missing file, openers with no gallery.js, the inline viewer put back), and (R-912) a catalog app's logo under another name than its slug — each must convict WITH its own failure line; and (2026-10-11, both of them reported from a real phone AFTER they went live) site.css's old mobile nav reaching the shell's sidebar from inside an @media block — the leak the STOP-1 check missed because it skipped @media entirely — and a <footer> left outside <main class=\"content\">, which on a shell page renders as a second column"),
+                        "answer, text after </html>, a Hungarian dashboard picture on an English page, and four picture-viewer faults (an opener without href, an opener to a missing file, openers with no gallery.js, the inline viewer put back), and (R-912) a catalog app's logo under another name than its slug — each must convict WITH its own failure line; and (2026-10-11, both of them reported from a real phone AFTER they went live) site.css's old mobile nav reaching the shell's sidebar from inside an @media block — the leak the STOP-1 check missed because it skipped @media entirely — and a <footer> left outside <main class=\"content\">, which on a shell page renders as a second column; and (2026-10-12) a sidebar #fragment pointing at an id no page carries - the 100 dead category links the operator found by clicking one"),
     "script-tests":    ("R-885: FIVE cases in scripts/test_script_tests_gate.py, run from here: a suite that PRINTS "
                         "OK and exits 1 (label without fact), a failing suite three levels deep (scope is a walk), "
                         "an empty scripts/ tree (checked nothing), the genuine article (must pass), the nested mark"),
@@ -657,6 +657,14 @@ decoy("site/shell-css-leak-in-media", "site_gates.py",
 #    shell page, so a <footer> left outside <main class="content"> is a flex SIBLING of the sidebar
 #    and the content, not the bottom of the page. The decoy closes <main> early and leaves the footer
 #    outside it — the shape the insert produced on every page that had a <script> after </main>.
+# 3. A DEAD #FRAGMENT IN THE SIDEBAR, which is copied into every page - so one wrong fragment is
+#    wrong twenty times. The five app-category links shipped pointing at ids nobody had written
+#    (#fajlkezeles-irodai for a section whose id is #fajlkezeles): 100 dead links, live, found by the
+#    operator clicking one. A dead fragment raises no error anywhere - the browser just stays put.
+decoy("site/dead-fragment", "site_gates.py",
+      replace_bytes(os.path.join(_WEB, "technologiak.html"),
+                    [('<li><a href="/alkalmazasok#media">', '<li><a href="/alkalmazasok#media-szorakozas">')]),
+      must='link to /alkalmazasok#media-szorakozas finds no id="media-szorakozas"')
 decoy("site/footer-outside-main", "site_gates.py",
       replace_bytes(os.path.join(_WEB, "gyik.html"),
                     [("    <footer>\n", "    </main>\n    <footer>\n"),