Download websites that use RequireJS

About RequireJS

The name suggests a small helper. What RequireJS actually shipped was a dependency resolver that ran in the browser at page load, reading declared dependencies, fetching each file only when something asked for it and doing all of that without a build step. The sites still carrying the loader form a coherent group rather than a random remnant.

RequireJS solved a problem that browsers eventually solved themselves. Released in 2009 under the MIT licence, it implemented asynchronous module definition so that JavaScript code could be split into modules with declared dependencies and fetched on demand, at a time when the alternative was a long list of script tags in a carefully maintained order. Native module support and modern bundlers made the approach unnecessary for new work, and the library settled into maintenance.

As of October 2026 we still see the loader on 179,425 sites, 45,269 of them in United States, with .com dominating the domain endings. Survival on that scale is not an accident. Front end dependencies are rarely removed, only buried, and a module loader sits so close to the foundation of a page that replacing it means reworking the build. Most of what remains lives inside enterprise portals, admin interfaces, university systems and vendor products that were delivered once and have been maintained rather than developed since.

For a modernisation agency, a list of domains still loading it is a list of codebases with deferred work and an owner who eventually has to face the bill. For a security team, an ageing loader is a reasonable proxy for an ageing dependency tree around it. For a vendor selling legacy replacement, it identifies accounts where half the argument is already made. None of those uses survive a per domain lookup, because the value sits in the size and the composition of the set rather than in any single verdict.

Detection depends on the file being fetched by the page we crawled. A build that inlines or renames the loader, a module system used only behind a login, or a page whose scripts did not execute during the crawl will leave a site uncounted, so the figure is a lower bound. Presence does not prove active use either, since a script tag can survive in a template long after the code that needed it was deleted. What the dataset supports is a dated statement about what each domain's visible front end loaded at crawl time, which is a more honest thing to build on than an estimate of what might be hidden.