Awesome-omni-skill linuse

Control the Linux/GNOME/Wayland desktop — screenshots, mouse taps, keyboard input. Use when asked to interact with the desktop, automate GUI apps, or do computer-use on the local Linux machine.

install

source · Clone the upstream repo

git clone https://github.com/diegosouzapw/awesome-omni-skill

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/diegosouzapw/awesome-omni-skill "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/development/linuse" ~/.claude/skills/diegosouzapw-awesome-omni-skill-linuse && rm -rf "$T"

manifest: skills/development/linuse/SKILL.md

source content

LinUse — Linux Desktop Automation (Wayland/GNOME)

Control the local GNOME Wayland desktop: take screenshots, tap/click, type text, press keys. Uses evdev virtual touchscreen + keyboard to bypass Wayland's input restrictions.

Prerequisites

GNOME on Wayland (tested on GNOME 44–46, Ubuntu 22.04–24.04)

gnome-screenshot

installed (

sudo apt install gnome-screenshot

)

```
evdev
```
Python package (
```
pip install evdev
```
)
```
/dev/uinput
```
writable (
```
sudo chmod 666 /dev/uinput
```
)

Screen lock disabled for automation:

gsettings set org.gnome.desktop.screensaver lock-enabled false
gsettings set org.gnome.desktop.session idle-delay 0
gsettings set org.gnome.desktop.interface enable-hot-corners false

Location

/path/to/linuse/

Quick Reference

All commands via:

cd /path/to/linuse && python3 -m linuse <command>

Single Commands

# Display info
python3 -m linuse info

# Screenshot (returns file path)
python3 -m linuse screenshot
python3 -m linuse screenshot -o /tmp/screen.png

# Tap at coordinates
python3 -m linuse tap 640 400

# Double-tap
python3 -m linuse double-tap 640 400

# Long press
python3 -m linuse long-press 640 400 --duration 0.5

# Drag
python3 -m linuse drag 100 100 500 500 --duration 0.3

# Type text (character by character via evdev)
python3 -m linuse type "hello world"

# Press single key
python3 -m linuse key Return
python3 -m linuse key Tab
python3 -m linuse key Escape

# Key combo
python3 -m linuse combo ctrl c
python3 -m linuse combo alt F4
python3 -m linuse combo ctrl shift t

Chained Commands (IMPORTANT)

For keyboard input to work reliably, chain commands in a single invocation using

as separator. This keeps the evdev devices alive across actions:

python3 -m linuse tap 640 60 , sleep 0.5 , type "example.com" , sleep 0.3 , key Return
python3 -m linuse combo ctrl l , sleep 0.5 , type "google.com" , key Return
python3 -m linuse combo alt F4

Why? Mutter registers/deregisters evdev devices when they're created/destroyed. Single commands create→use→destroy per invocation. Chaining keeps them alive.

Computer-Use Workflow

1. Screenshot  →  python3 -m linuse screenshot -o /tmp/screen.png
                  → analyze with image tool
2. Decide      →  identify target coordinates or text to type
3. Act         →  python3 -m linuse tap X Y , sleep 0.3 , type "text" , key Return
4. Verify      →  python3 -m linuse screenshot -o /tmp/after.png
                  → analyze with image tool
5. Repeat

Example: Open a URL in Firefox

# Focus address bar, type URL, navigate
python3 -m linuse combo ctrl l , sleep 0.5 , type "example.com" , sleep 0.3 , key Return
sleep 3
python3 -m linuse screenshot -o /tmp/result.png

Example: Close current window

python3 -m linuse combo alt F4

Example: Type in a text field after clicking it

python3 -m linuse tap 400 300 , sleep 0.3 , type "Hello from Clawdia"

Coordinate Tips

Screen resolution detected automatically (check with
```
info
```
)
Coordinates are in logical pixels (1280x800 on this machine)
Insert
```
sleep 0.3
```
between tap and type to let focus settle
Insert
```
sleep 0.1
```
between screenshot and the previous action
For dock icons: x≈27, y varies by icon position (top of dock ≈ 150-170)

Limitations

Touch semantics only — no hover state, no right-click (it's a touchscreen)
No clipboard paste — type goes character by character (slow for long text)
No window enumeration — can't list windows like winuse does
GNOME-specific — won't work on KDE/Sway without changes
Needs uinput access — requires chmod or udev rule

Troubleshooting

Keyboard input not reaching app

Use chained commands (
```
,
```
separator) — single-shot keyboard commands may not work
Make sure the target window has focus (tap it first, add sleep)

Screenshot returns black

Screen might be locked:
```
loginctl unlock-session
```

Display might be off:

gsettings set org.gnome.desktop.session idle-delay 0

"Permission denied" on /dev/uinput

sudo chmod 666 /dev/uinput
# Or persistent: echo 'KERNEL=="uinput", MODE="0666"' | sudo tee /etc/udev/rules.d/99-uinput.rules

Taps land in wrong place

Verify resolution with
```
python3 -m linuse info
```
Take a screenshot and estimate coordinates from the image
Remember: coordinates are absolute screen pixels