Chrome 154 stabilized on September 22, 2026, and the release quietly closes one of the more awkward gaps left by scroll-marker-group: the carousel dots and horizontal-scroll tabs that sites have been building with the property since it shipped had no agreed-upon keyboard or assistive-technology behavior. That changes with this release, alongside a new text-decoration control and a long-requested fix for iframe sizing — the kind of unglamorous plumbing that standards-driven frontend engineering depends on more than any single flashy API.
Three changes matter most for people building production interfaces:
links treats them like a set of anchor links, tabs treats them like a tabbed interface following the WAI-ARIA tabs pattern.postMessage height-sync hack that's been standard practice for a decade.Before this release, a ::scroll-marker-group built for a horizontally scrolling gallery looked identical, in the accessibility tree, to one built for a tabbed panel switcher — because the browser had no way to know which one you meant. Screen reader users got generic "list" semantics regardless of the actual interaction pattern, and keyboard focus order was whatever the DOM happened to produce. The new links and tabs keywords fix that at the CSS layer: declare the interaction pattern once, and the browser assigns the matching roles and focus behavior automatically. That's a meaningful reduction in the custom ARIA wiring teams previously had to hand-roll for anything built on scroll-driven navigation.
The property takes one or two length values, following the same syntax shape as inset(): positive values pull a decoration in from the text edges, negative values push it past them. That second case is the interesting one — it's what makes an underline "reveal" animation on hover possible with a single declaration instead of an absolutely positioned pseudo-element faking the effect.
Embedded widgets, payment forms, and third-party components that live in an <iframe> have always had an awkward sizing problem: the parent page doesn't know the child document's real height until the child tells it, usually via a postMessage round-trip wired up by hand on both sides. Chrome 154 lets the embedded document report its layout overflow size directly, so the parent frame can size itself to match without any JavaScript bridge. It's a small API with an outsized effect on how much glue code embed widgets need to ship.
None of these three features require a framework upgrade or a build step — they're CSS and platform primitives you can adopt incrementally. If your product has anything resembling a horizontally scrolling tab strip, card carousel, or story-style navigation, scroll-marker-group's new modes are worth auditing against today, because the accessibility gap they close was previously invisible in casual testing and only showed up with a screen reader or keyboard-only navigation. Teams maintaining embedded widgets should treat the native iframe sizing API as a deletion opportunity: once support is broad enough, the postMessage height-sync code becomes dead weight.
Chrome 154 is stable as of September 22, 2026. As with most CSS additions, cross-browser support follows on its own schedule — check the feature's entry in the WHATWG HTML Standard ecosystem and MDN before shipping any of these as a hard dependency, and keep a progressive-enhancement fallback (plain focus order, a manual underline, or a fixed iframe height) for browsers that haven't caught up yet. None of the three features change existing behavior for pages that don't opt in, so there's no migration risk in waiting.