wt-stack.js (measure fixed bars → set body padding + --wt-stack-h, promo-bar z 2147483647→1000) injected on all 38 content-hidden sites (~6,600 pages, per-site backups in /root/site-snapshots/); the two true bar-on-bar overlaps (cargosolar ticker, cargotown subnav) got exact CSS offsets. Regression re-audit below: 0 overlaps, 0 content-hidden, 0 sites on z-index 2147483647 (was 38 High / 2 overlaps / 165 max-z). The max-int z-index was retired at the source on 2026-07-02: 18 /root generator scripts + 13,975 static pages across 178 sites now use z:1000 for network chrome (backup /root/zfix-scripts-backup-260702.tgz, change map /root/zfix-changed-260702.json). Injector: /root/wt-stack/inject-wt-stack.py (has --remove undo).Every public site rendered live in headless Chromium (1280×900) and measured for competing fixed/sticky top bars: overlaps, stacked height, whether page content clears the bars, and z-index. This is what the pages actually do in a browser — not a guess from the HTML.
Over time the network accumulated several independently-injected bars that each assume they own the top of the page: the #promo-bar ("WholeTech Picks", on 169 sites, position:fixed; top:0; z-index:2147483647 — the maximum 32-bit value), a #wholetech-back link (58 sites, top:6px — literally inside the promo-bar's band), a #subnav-bar (47 sites, top:52px), each site's own .nav (108 sites, many position:sticky; top:0), plus an in-flow af-strip (130 sites) and vid-teaser (64 sites). The promo-bar ships a JavaScript routine that tries to push every other fixed/sticky element down and pad the body — but it runs once, races late-injected bars and sticky navs, and when it miscalculates, two opaque bars land in the same vertical space (overlap) or the body padding is too small and real content hides behind the stack. The table below flags exactly which sites that happens on.
body padding clear the pinned bars, or is content hidden underneath? (No = hidden.)2147483647 is the promo-bar pinning itself above everything, including legitimate site modals.Click any row for the per-bar breakdown and a screenshot of that site's top band.
Replace the N independent fixed bars + the fragile per-bar offset script with a single owner of the top edge:
A single #wt-stack at top:0 holds the network chrome (promo picks, back-link, subnav) as stacked children that flow vertically — so they can never overlap each other. One element, one z-index.
A ResizeObserver on #wt-stack sets --wt-stack-h on :root. The body uses padding-top:var(--wt-stack-h) and scroll-padding-top:var(--wt-stack-h). Content always clears the stack, at any width, automatically — no per-element scanning, no race.
Each site's sticky .nav uses top:var(--wt-stack-h) instead of top:0, so it pins directly under the network chrome instead of fighting it for top:0.
Retire z-index:2147483647. Use a documented scale: stack 1000, site nav 900, site modals 2000 (so a site's own dialogs can sit above the promo chrome when needed).
Collapse the separate promo-bar / back-link / subnav injectors into a single idempotent rollout, then re-run this exact Playwright audit to confirm 0 overlaps and Content-clear = Yes network-wide before and after. The worksheet becomes the regression test.
| # | Site | Severity | Bars | Overlaps | Stacked px | Content clear? | Max z | Top bars |
|---|---|---|---|---|---|---|---|---|
| Loading audit data… | ||||||||