Download websites that use Mura CMS

About Mura CMS

Almost everything about this technology follows from the runtime underneath it. Mura CMS is written in ColdFusion, and a CFML application implies an engine to execute it, which may be a commercially licensed product with a real per-server cost. Nobody arrives at that stack casually. The organisations running it either standardised on it long ago and built around it, or inherited it with an application they could not afford to rewrite.

Which is why the detection carries more information than the site count suggests. As of September 2026 we detect the platform on 558 sites in the Blogs category, and behind each of those is an infrastructure decision with budget attached. The adopter profile is consistent: universities, government departments, healthcare organisations and mid-sized enterprises with long-lived intranets and public sites, sectors where continuity outweighs fashion and where a working system is kept working for a decade at a time.

Version data makes that inertia measurable rather than assumed. We currently observe 66 distinct version strings across the detected population, with 7.0 the release reported by the largest number of sites. A spread that wide across a population this small is the signature of an estate maintained in place rather than upgraded on a cadence. Sites sit on whatever release they were delivered on, and they move only when something forces them to move.

For a commercial reader the fragmentation is the product. A migration practice looking for rebuild work wants the older releases. A hosting provider wants to know which engines and operating systems sit underneath them. A security team wants the releases that no longer receive fixes. Each of those is the same population filtered differently, which is a query you can only run against a full list held locally.

Blueriver has developed the platform since 2007. Our index records the licence as GPL v2, and the source published for the core does not bear that out, because its files carry the copyleft GNU General Public License instead. Open source the codebase certainly was, which is why it later continued under a community fork as well. The fork matters for reading the data. A forked descendant shares most of its markup fingerprints with its ancestor, so detections near the fork boundary can be ambiguous and some sites recorded here may in practice be running the successor project. We would rather say so than present a clean number we cannot fully defend.

The remaining caveats are the ordinary ones. Identification comes from public markup patterns, asset paths and headers, and a site that has been heavily themed, placed behind a caching proxy, or configured to suppress identifying headers may not be attributed at all. Version strings appear only where the application discloses them, so the versioned subset is smaller than the detected subset and skews toward default configurations.

What the export gives you is every domain where the platform was recognised, with the observed version, the hosting country, the domain zone and the rest of the stack on the same row, rebuilt daily and filtered on your own machine.