review round 2: deterministic checks (npm run check: build + zero-dep link checker; tsconfig with docs-recommended include/exclude, astro check dropped — OOMs at 12GB on a 200-entry collection), sitemap static pages derived from route files, non-recursive content glob (id/slug trap), --header-h token, honest CDN comments (sveltia UI fonts), preview.css drift (prose a wrap, img block, h1 size)

src/pages/news/[slug].astro

@@ -122,7 +122,8 @@ const readMin = readMinOf(post);
     let observer;
     const observe = () => {
       observer?.disconnect();
-      // Narrow band below the sticky header (56px) — Starlight's rootMargin trick
+      // Narrow band below the sticky header (--header-h 56px + 12px slack;
+      // JS can't read the token cheaply, keep in sync) — Starlight's rootMargin trick
       const top = 68, band = 64;
       observer = new IntersectionObserver(onIntersect, {
         rootMargin: `-${top}px 0px ${top + band - document.documentElement.clientHeight}px`,

src/pages/sitemap.xml.js

@@ -13,10 +13,17 @@ export async function GET(context) {
     .sort((a, b) => b.data.date.valueOf() - a.data.date.valueOf());
   const pages = await getCollection('pages');
 
-  // Go-live note: /login and /dashboard are mock pages — drop them from this
-  // list (or gate them) before robots.txt stops disallowing everything.
+  // Static pages come from the actual route files, so a new src/pages/*.astro
+  // can never be forgotten here. 404 is excluded; /admin (subdirectory) is
+  // intentionally unlisted. Go-live note: /login and /dashboard are mock
+  // pages — filter them out before robots.txt stops disallowing everything.
+  const staticPages = Object.keys(import.meta.glob('./*.astro'))
+    .map((p) => p.slice(1, -'.astro'.length))
+    .filter((p) => p !== '/404')
+    .map((p) => (p === '/index' ? '/' : p))
+    .sort();
   const urls = [
-    '/', '/news', '/proposals', '/dashboard', '/support', '/search', '/login',
+    ...staticPages, '/news',
     ...Array.from({ length: newsPageCount(news.length) - 1 }, (_, i) => `/news/${i + 2}`),
     ...pages.map((p) => `/pages/${p.id}`),
   ].map((path) => `  <url><loc>${new URL(path, site).href}</loc></url>`);