Download websites that use GitHub Pages

Publishing through GitHub Pages rules things out as much as it rules things in. There is no application server behind the domain, no database, no server side session, and no administrative login. Whatever a visitor receives was built somewhere else, committed to a repository, and copied onto a static file service. Presence of GitHub Pages on a domain says less about what the site does than about how the people who own it work, which is version control first and deploys as commits.

The population that results is unusual for a PaaS entry. Open source project sites, developer portfolios, library documentation, conference microsites and internal tooling front ends all cluster here, because all of them are cheap to render ahead of time and all of them already live in a repository. Commercial sites appear too, but usually as satellites: a launch page, a status page, a docs subdomain hanging off a company that runs its real product elsewhere. Anyone selling developer tools, documentation platforms, package registries or continuous integration is looking at a directory of teams that have already accepted a git centred workflow, which is a shorter conversation than starting from nothing.

As of October 2026 we detect GitHub Pages on 1,135,724 sites, and 1,089,411 of those resolve to United States. The country concentration deserves a warning rather than a headline. A centralised service answers from its own infrastructure, so geolocation describes where the platform terminates the request and not where the publisher sits. A country column on a centrally hosted service is a fact about the provider before it is a fact about anybody's market. The top level domain is the more honest clue to audience, and for these sites that clue is .com.

GitHub (Microsoft) has run the service since 2008 and keeps it Proprietary, offered without cost for public repositories along with custom domains, encryption and build integration. A directory of these domains is therefore volatile at the edges. Projects go quiet, redirect to a documentation host, or graduate to real hosting once they need forms and accounts. Churn is itself the interesting variable if you are tracking where small projects go when they outgrow static files. Zero cost is also why a good part of the population is abandoned rather than active. A repository that stopped receiving commits keeps serving its last build indefinitely, so a live domain here is not proof of a live project.

A per domain lookup answers one question, which is whether one site is static and repository published. The downloadable list answers the harder inverse question of who all of them are. Filter by country, by suffix, by the other technologies detected alongside, and you have a segment rather than an anecdote: documentation sites that also load a search widget, portfolios carrying an analytics tag, project pages still pointing at a dead custom domain. Detection here is honest but not exhaustive. A domain fronted by a proxy that hides its origin will not always report the service underneath, so the figure should be read as sites we can confirm rather than every site in existence.