The sources for https://browser-specs.github.io/: a list of every browser spec in browser-specs, one row per level — showing which engines implement it, its WPT pass rates, its MDN articles and Can I Use features, and its “platform footprint” (how many other specs build on it).
The list starts with the most-recently-updated specs, and sorts by any of its columns. Its Filter box finds specs by title, shortname, org, or group, and a word in it that starts with - hides the specs that match that word (-css). A spec’s name opens a panel that says what the spec is — its status, abstract, editors, groups, and other levels — and its counts, test bars, and logos open panels with the details.
None of the data is hand-maintained, apart from a few display names in src/logic/rows.ts. Every hour, a GitHub Actions workflow builds the site from browser-specs, webref’s crawl (for each spec’s date), web-features (for its categories), and the data files Web Specs Companion builds from BCD, mdn/content, caniuse-db, and wpt.fyi, and deploys it when anything has changed.
A spec’s “platform footprint” comes from webref’s backrefs, links, refs, and idlparsed extracts for it, fetched on every build. A spec’s abstract and editors come from the start of its page, read on every build, which asks each server only for a change since the copy the last build deployed (and from Specref, for a page that names no editors). And a spec’s category is a guess, made by src/logic/categories.ts — from what those sources say about it.
Needs Node 24. npm ci, then npm run build writes the site into _site/, and python3 -m http.server -d _site serves it. npm test runs the tests, and npm run yardstick checks the category guesses against the old hand-kept list.
CC0 1.0. The browser logos in site/icons/browsers/ are from alrra/browser-logos, under the MIT license in LICENSE.txt beside them; the logos themselves are trademarks of Mozilla, Apple, and Google.