claw-mouse
This skill provides a small, scriptable desktop GUI control helper for Linux X11.
It’s intended for “vision loop” automation:
- take a screenshot
- decide where to click
- move/click/type
- repeat
Under the hood it wraps:
scrotfor screenshotsxdotoolfor mouse/keyboard/window control
Files
desktopctl.py— the CLI script
Requirements
- Linux running X11 (not Wayland-only)
python3xdotoolscrot
Ubuntu/Debian:
sudo apt-get update
sudo apt-get install -y xdotool scrot
Usage
From this skill directory:
python3 desktopctl.py screenshot
python3 desktopctl.py click 500 300
python3 desktopctl.py type "hello"
python3 desktopctl.py key ctrl+l
python3 desktopctl.py windows
python3 desktopctl.py activate "Chromium"
DISPLAY / XAUTHORITY
If you’re running from a daemon/headless shell where DISPLAY isn’t set:
DISPLAY=:0 XAUTHORITY=$HOME/.Xauthority python3 desktopctl.py screenshot
Or use flags:
python3 desktopctl.py --display :0 --xauthority $HOME/.Xauthority screenshot
Safety notes
This can click/type into your real desktop session. Use carefully.
Changelog
- 0.1.0: Initial published skill.