It is an admin tool with a moodle cron task, but it reaches into your moodle via curl effectively from outside moodle, and scrapes each page, parses it and follows links. By using this architecture it will only find broken links that actually matter to students. Because it comes in from outside it needs to authenticate and has a dependancy on the moodle-auth_basic plugin. It is recommended that you setup a dedicated 'robot' user who has readonly access to all the site pages you wish to crawl. You should give the robot similar capabilites that real students will have.