Download websites that use Animate.css

About Animate.css

A library of prewritten keyframes has no real reason to churn, and yet we record 399 distinct version strings for Animate.css across the sites that carry it. Fragmentation on that scale is not a sign of rapid development. Published in 2013 by Daniel Eden under the MIT license, the project has shipped a modest number of genuine releases in its lifetime. What the spread reflects is the way the file arrives on a site: copied into a theme, vendored into a page builder, pasted out of a tutorial, and then never touched again.

Because nobody upgrades a decorative dependency, the version frozen into a page is usually the version that was current on the day the front end was assembled. A large share of installs report 4.1.1, and the long tail behind that value is effectively a sediment layer of older builds. The column therefore behaves as a date stamp rather than as a support matrix. A site running an ancient copy is not exposed to anything dangerous, since the file contains no logic at all, but the copy does tell you when somebody last rebuilt the presentation layer.

As of September 2026 we detect Animate.css on 2,750,083 sites, filed under UI frameworks. United States contributes 1,011,089 of them and .com carries much of the volume, a distribution that follows the reach of the commercial theme marketplaces more closely than it follows anything happening in front-end engineering.

For a researcher the practical implication is about provenance, not about animation. A page that loads this stylesheet was in most cases assembled from a template, a builder, or an agency starter kit rather than written from scratch by a front-end team. Vendors selling into the template ecosystem use presence as a qualifier. Agencies pitching redesigns use the frozen version as an age estimate. Performance consultants use it as a marker of unused CSS, because a site typically ships the entire animation set in order to use a handful of its classes, and the whole file sits in the critical path for the sake of a fade.

A single lookup tells you one site loads a decorative stylesheet, which decides nothing. The downloadable list turns the same field into a cohort: every site on a given version, grouped by country, by domain suffix, and by whatever else sits in the same record. Grouping is how you separate the builder populations from the hand-rolled ones, how you date a market segment, and how you size it before spending anything to reach it. None of those operations exist in a lookup box that answers one domain at a time.

The detection has real limits and they run in one direction. When a build pipeline concatenates and minifies the stylesheet into a bundle, the header comment that carries the version disappears and the file often stops being recognizable as itself, so our count is a floor rather than a total. Copies renamed inside a theme are missed for the same reason. The version we publish is the one the file declares about itself, which is trustworthy when present and simply absent when a toolchain has stripped it. Anyone building an argument on the version column should treat missing values as unknown rather than as old.