Download websites that use Chart.js

About Chart.js

A charting library says something about a page that most front end dependencies do not. Chart.js does not arrive with a CMS and no hosting company installs it by default. Someone added it because the page had to display numbers, which usually means a dashboard, a report, a pricing calculator, a status view or a public data set sitting behind the interface.

Released under MIT and in circulation since 2013, the library draws into a canvas element rather than into markup. Drawing into a canvas keeps the payload small and keeps the output opaque to anything that only reads the document tree, which is worth knowing if you plan to scrape what the charts contain rather than simply record that charts exist.

As of September 2026 we detect the library on 331,883 sites. 127,245 of those are served from United States and .com runs through much of the suffix mix, but geography is the less interesting axis for a dependency that travels inside templates and admin panels rather than inside a national market.

Version spread is the axis that pays. We record 91 distinct versions in the field, and 2.9.4 remains widely present even though the project has moved on through later major lines. A page pinned to an older major line is usually a page nobody has rebuilt, because charting code is the sort of dependency that gets touched only when it breaks or when the whole front end is replaced. For anyone selling a component library, an analytics product or front end modernisation work, the split between current and stale builds is the segmentation that decides who is worth approaching.

A per domain lookup tells you whether a company you already had in mind draws its charts this way. A downloadable file answers the opposite question. Filter it by version and by country and it returns the population of sites that render data in the browser, which is a workable proxy for sites with a reporting surface, an internal tool exposed to customers or a public metrics page. That population is not something you can assemble by checking domains one at a time, because you would need to know the answer before you knew where to look.

Detection depends on the library identifying itself in what reaches the browser. A build that inlines and minifies everything into a single application bundle can strip the recognisable marker, so applications shipped through a modern bundler are underrepresented compared with sites that load the library from a script tag or a shared CDN. The count is a floor, and the bias runs toward simpler deployments, which is the honest way to read any row in the file.