Click an element or screen coordinate. Delivers a real mouseDown/mouseUp pair via process-routed SkyLight events for custom-drawn controls, or AXPress/AXPick/AXConfirm accessibility actions for standard controls.

Drag from one element or coordinate to another. Delivers mouseDown, mouseMoved, and mouseUp events via process-routed SkyLight events.

Get the accessibility tree and screenshot of an app window. Must be called before any action on an app, because it produces the element_index handles actions refer to.

Move the cursor to an element or coordinate without clicking. Does not move the actual system cursor; only the agent's internal tracking.

Every running application with a GUI presence. Ordered so the frontmost app comes first: when a user says "the browser" or gives an ambiguous name, what they are looking at is the best prior.

Perform a secondary action on an element, such as AXShowMenu for context menus or other custom accessibility actions.

Press a key or key combination. Delivers keystrokes via process-routed SkyLight events.