Your Mouse Got Sick and You Don’t Know it. aka “Reverse Shell via Mouse”

Ever got a backdoor installed on your computer by your beloved mouse? Here’s the story of a poor mouse that got really, really sick.

Agent "Danger Mouse"
Agent “Danger Mouse”

Do you remember the times where people put Teensy-boards and USB hubs in their mouses? [Chris? ;)] Their aim was to attach an additional Human Interface Device (HID, like keyboards or mouses) with some payload in kind of e.g. keystrokes or mouse movements. Also, there are devices available like the USB Rubber Ducky in the housing of a USB thumb drive.
The principle is easy: The tools are using a programmable microcontroller with the capability to emulate USB HID. That’s it. Just program your board of choice with the payload fitting your needs and plug it in at the target computer. The latter will recognize it as a keyboard/mouse and the payload-keystrokes will be entered.
But why should external hardware be used? Many modern gaming peripherals provide functions to store macros on them, including enough onboard memory for little payloads.

While we were working on the layout of this years’ Troopers-Badge, I felt uncomfortable using my touchpad, so I switched over to a Logitech G-series gaming mouse. This one worked like a charm – many buttons and the feature to store personal profiles on the mouse itself, which is perfect when you work on more than one machine.

But wait – macros and profiles stored on the mouse? Recall the lines above concerning the HID story.
Could it be possible to store a macro big enough to drop a reverse shell on a Windows target?
Actually – it could.
It’s just as simple as using the Logitech Gaming Software’s Command Editor. Choose a button, put a macro on it, fit the timings and go!
The only thing you should consider, that you’re limited to about 100 keystrokes. If there should be something dropped on the target, like an executable or a script, you should think about using FTP or Powershell to download it externally, like I did here.

In this Proof of Concept the marco opens the Windows Command Line and downloads Netcat via Windows’ own ftp.exe from an external FTP server. Afterwards, it launches Netcat in background mode, while a Netcat listener already is waiting on the remote machine.

I think this kind of attack is dangerous due to its minimal effort and people trusting their mouses. Who would be afraid of a manipulated mouse, while being away for just 2 minutes to grab a cold Mate drink in the kitchen? The mouse’s software in this example offers the option to delay the keystrokes. Consider, the mouse acts like usual and after 30 or 60 seconds the bad magic happens, and it would take only a few seconds, so you might even miss the chance to see the windows popping up.


Live long and prosper,