What makes Android touch automation predictable across devices?

Building an Android auto-clicker taught me that dispatching a gesture is the easy part. The harder problem is making every action understandable and predictable for the user.

For Phone Operator, I have been using four design rules:

1. The marker must show the real action. A tap marker should display the configured random radius, while a swipe marker should show direction and distance without moving during execution.

2. Every step needs independent settings. Timing, delay, repeat count, random range, and swipe distance should belong to the individual action instead of a global template.

3. The controller should stay out of the way. It must remain movable and transparent across supported apps, but its position should not change when a swipe runs.

4. Permissions need plain-language boundaries. AccessibilityService only executes gestures configured and started by the user; it does not read passwords or collect screen content.

The app supports taps, double taps, four-direction swipes, delays, loops, and mixed reusable scripts without root. I am testing on Samsung and would value practical feedback from Xiaomi, Huawei, Pixel, and other Android users.

Which matters most for trust in an automation tool: visible action ranges, per-step timing controls, a run preview, or clearer device-specific permission guidance?