Download websites that use MkDocs

About MkDocs

Much of what we file under Documentation exists to help non-technical people publish. MkDocs exists to help engineers publish documentation, and the difference shows up in every part of its shape. Released in 2014 under the BSD licence and written in Python, the generator turns a directory of Markdown files and a single YAML config into a static documentation site. As of September 2026 we see it on 7,054 sites.

A site built this way has no database, no login and no editorial workflow. Content lives in a repository beside the code it describes, and publishing happens when a pipeline runs. Documentation therefore versions with the software, changes go through code review, and the people writing the pages are the people who wrote the product. Any site we detect running this generator is, with high confidence, a site owned by a software team.

Version data reflects that ownership. We record 39 distinct releases in circulation and see 1.6.1 more often than any other, and the concentration is tighter than it would be for a consumer publishing tool. Sites regenerated by continuous integration pick up whatever the pinned requirement resolves to at build time, so the version distribution is closer to a picture of dependency hygiene than of manual upgrading. A stale build on a documentation site usually means a stale pipeline rather than an absent administrator.

The commercial reading is straightforward. A list of documentation sites is a list of companies that ship software and are far enough along to document it. Developer tool vendors, API platforms, documentation hosting services, technical writing agencies and search products aimed at docs all want that population, and it is difficult to assemble any other way, because these sites are scattered across corporate subdomains and project domains that no directory collects.

Detection depends on markers the generated output leaves behind, chiefly generator metadata and the asset structure of the default and Material themes. A team that strips metadata, serves the output through a heavily customised theme or wraps the docs behind a corporate proxy disappears from our view entirely, so the figure is a lower bound on real usage.

Looking up a single domain answers whether one company's docs run on this generator. The downloadable file answers which companies run documentation at all, on which domains and subdomains, on which hosts, and which of them changed generator since the previous export. A question about a population cannot be asked one domain at a time.