List interactive and labelled elements on screen with handles. Optionally attach a screenshot.

Current screen as JPEG, scaled to width px wide (the device is usually 1080 wide; scale coordinates back).

Tap an element handle ("e7") or device coordinates ("540,1200").

Tap the first element whose text (or content description) contains text.

Press and hold an element handle or "x,y".

Swipe from start to end (handles or "x,y"). Set hold_ms > 0 to press first, i.e. drag.

Scroll "down" (reveal content below), "up", "left" or "right", within a scrollable handle or the whole screen, by about distance of its height/width so consecutive pages overlap. fling=True lets the list coast further (faster for long lists, but items can be skipped).

Two-finger pinch: zoom "in" or "out", centred on a handle/"x,y" or the screen centre. Emulator gRPC only.

Turn the device to "portrait", "landscape", "reverse_portrait" or "reverse_landscape", like a person turning the phone. The display only follows if auto-rotate is on and the app allows it; check the orientation in the returned header.

Type into the focused field (tap the field first). Unicode is fine. clear=True replaces existing text.

Press a key: back, home, recent, delete, tab, volume_up, volume_down, power, or any KEYCODE_ name. "enter" (also "search"/"go"/"send"/"done") presses the on-screen keyboard's action key when the keyboard is shown, which is what submits most search boxes and chat inputs.

Wait until an element whose label (text or content description) contains text, and/or with resource id resource_id, appears.

Wait until the screen stops changing and nothing is loading (use when the element list says loading, e.g. a web page still coming in), then return the element list.

Launch an app by package name (see app_list).

Force-stop an app.

Foreground package and activity.

Installed packages, optionally filtered by substring.

Which backends are in use, screen size and pacing.