The problem
Every once and a while I need to create a click-through demo and capture switching between browsers/tabs/applications. Most click to capture apps only capture HTML and other code from webpages. This leaves you with the problem of having to record a video of any actions not in the browser and splice it into your demo (with all of the quality and resolution issues that come with that), or try and find a way to avoid doing it altogether.
My solution
Being a perfectionist, this really bothered me. I searched around for a solution, and found a docker container that hosted a React replica of Windows 11, but the functionality was pretty limited. I didn’t find much else otherwise.
Why not just sit down and use Qwen to build my own HTML desktop? It should be possible, right? I started up OpenCode and connected it to my local LLM machine, and in about an hour I had something pretty cool.
I added some features like a start menu, an iframe browser that you could actually use to browse (most) sites, and even a network file share explorer that would take filenames and file attributes from a textfile and show them in the explorer window like real files. This allows you to show ‘restored’ files if you wanted to show a file recovery after a breach or attack.
Our Navattic capture software will capture every element, since its HTML, and the items are even clickable inside the captured desktop, so you can click icons, change tabs and browse between websites.
Best of all, it looks like KDE and not like Windows.
Grab the files here and try them out. Github - HTML Desktop