What problem does it solve? Automating Linux desktop GUI applications is error-prone when agents must guess pixel coordinates or target the wrong window. This Skill provides a two-stage look-then-act workflow: inspect the accessibility tree to get exact widget coordinates, then actuate clicks, typing, and key presses via xdotool. ## Core Features & Use Cases - Accessibility-tree inspection: Dump the widget hierarchy of any window as JSON with role, name, and x/y/w/h coordinates, or find a single widget's bounding box in one command. - Coordinate-based actuation: Activate windows, click coordinates, type literal text, and send special keys through a thin xdotool wrapper. - Accessibility-aware launching: Launch Electron and Chromium apps with --force-renderer-accessibility so their widgets appear in the AT-SPI tree. - Use Case: Automate a Save As dialog in VS Code by inspecting the dialog tree, locating the Save button's coordinates, computing its center, and clicking it. ## Quick Start Use the linux-gui-control skill to open Chromium, navigate to a URL, and take a screenshot of the loaded page.
