Honestly, I started building this as a basic Python script with a libratbag backend that could support my mouse DPI switches and button remaps without needing two separate programs—and without the dang notifications being broken for those DPI switches.
After I started coding with AI more, mostly because some of this became very complicated and I just wanted something that worked, I discovered something by actually reading the HID reports: there is a missing piece of the puzzle when it comes to supporting mice on Linux.
When you make these devices work, you’re essentially piecing together HID bits and protocol behavior. So I started wondering: what if the program could understand those pieces itself, using a broad repertoire of knowledge that dozens of people have worked hard to uncover over the years?
That’s what I think I ended up making.
The engine essentially pieces together HID information and behavioral evidence to figure out how to get every possible function of a mouse working. It has become an insane program compared with what I originally intended to build.
At this point, I mostly need testers.
I’ve tried to improve the readability of the program, but I’ve also security-tested it and validated its functions into the ground. I’ve stayed up for hours pressing the same button over and over and over again, confirming that this isn’t a fluke or some coding lie.
This is something different.