Every agency I know has the same problem: dozens of client websites, built years ago, still running. Nobody remembers which PHP version is on which server, or whether that old WordPress theme still pulls in jQuery 1.12. The software keeps working, so nobody notices that it stopped getting security fixes years ago.

So I built Website EOL Risk Scanner, an Apify Actor that takes a list of websites and tells you which ones run end-of-life software.

What it does

For every website it reports the detected technologies (PHP, WordPress, Joomla, Drupal, TYPO3, Magento, nginx, Apache, jQuery, Bootstrap, AngularJS), checks each version against endoflife.date and returns:

  • a risk score from 0 to 100
  • the end-of-life date and the latest version of each technology
  • one plain sentence, e.g. "PHP 7.4.33 reached end of life on 2022-11-28 (3 years ago)"

How detection works

The scanner only reads what any browser receives: response headers (X-Powered-By, Server), the generator meta tag, and the URLs of scripts and stylesheets. That covers a lot, but not everything.

The surprise came when I ran it on applications I maintain myself. They are built on Yii 2, which publishes assets into hashed folders like /assets/3240b52c/jquery.js. No version anywhere in the URL, so the first version of the scanner reported jQuery and Bootstrap as "unknown".

The fix: with deep scan on, the Actor downloads up to two self-hosted library files and reads the licence banner at the top (Bootstrap v3.4.1). On my own apps that turned up Bootstrap 3, out of support since 2019. Nobody on the team had noticed.

"Supported" is not a yes/no question

endoflife.date tracks two dates per release line: end of active support and end of life. Old WordPress lines, for example, still get security patches long after active support ends. Calling them "safe" would be wrong; calling them "end of life" would be wrong too. The scanner reports them as security-only and gives them a smaller weight in the score.

A Crawlee gotcha with robots.txt

The scanner respects robots.txt. I first used crawler.addRequests() with respectRobotsTxtFile: true. On a list of 5,000 websites nothing happened for a very long time: addRequests() fetches robots.txt for every URL one after another before the first request is processed. Filling the RequestQueue directly and letting the crawler check robots.txt per request, in parallel, fixed it.

What it found on 8,000 websites

To see how common this is, I ran the scanner on two samples from the Tranco list: the 5,000 top-ranked Czech (.cz) domains and 5,000 domains spread across ranks 10,000 to 1,000,000 worldwide. 8,358 of them could be scanned.

  • 21% of the Czech sites and 17% of the global ones visibly run at least one end-of-life technology.
  • 14% of the Czech sites still load jQuery 1.x or 2.x, out of support since 2016 (9% globally).
  • Of the sites that reveal their PHP version, about half run an unsupported one. 110 still run PHP 5 or older.
  • About 4% run an end-of-life PHP, CMS or Apache version.

These are lower bounds: most sites hide their versions, and the scanner only counts what is visible. nginx and OpenSSL versions are left out because Linux distributions often patch them themselves.

Try it

Website EOL Risk Scanner on Apify Store

It costs $8 per 1,000 scanned websites; websites that fail to load are free. I would love to hear which technologies you want added next.