Injecting Mouse and Keyboard — performMouseInput, performKeyInput, performMultiModalInput

Mouse and keyboard input are first-class modalities in Compose tests. Each has its own injection scope and entry point, plus a shared performMultiModalInput that combines all of them inside one batched gesture (touch, mouse, key, rotary, trackpad, indirect pointer). Coordinates remain node-local — the same rules as touch — and the entire injection state survives across perform.*Input blocks on the same node.

When to use this skill

• The test exercises a desktop or Compose Multiplatform UI that responds to clicks, hover, right-click, scroll wheel, or modifier keys.
• The test must drive a keyboard shortcut (e.g. Ctrl+S, Shift+Tab, arrow-key navigation).
• The test must verify hover-only behaviour: tooltip pop, color change on Modifier.hoverable, mouse-only ripple state.
• The test simulates a drag-and-drop with a mouse pointer.
• The test must combine modalities — for example "hover at (x, y) then click and hold while pressing Shift".

When NOT to use this skill

• The interaction is a finger tap or drag — use ../injecting-touch-gestures/SKILL.md. Mouse is not interchangeable with touch on Android device tests.
• A simple performClick() already routes through the right modality per platform — use ../clicking-and-scrolling/SKILL.md.
• The test types into a TextField — use ../entering-text/SKILL.md. Key input here is for shortcuts, not text entry.
• The test waits for a hover-driven animation to settle — pause the clock first per ../../synchronization/testing-animations-deterministically/SKILL.md.