This app is written in vanilla JS, HTML and CSS. It uses a php backend, with file system storage. On the frontend the user identity is generated from looking at the extensions, ram, cpu, screen size, dpi and color scheme preferences. Additionally a canvas is used to paint some fonts and shapes, which is saved as a PNG from which the image checksum is used. Click this card to go to the Browserleaks website for an explanation for how this works. These markers are used to produce a client has, which is matched on the server, against user agent and ip information. Here a tensor flow model is used to allow some leeway in changing markers like IP, plugins, etc.