dockkit

Framework guidance for DockKit, which integrates iOS apps with motorized camera stands and gimbals that physically track subjects by rotating the iPhone. DockKit handles motor control, subject detection, and framing so any app using AVFoundation camera APIs gets 360-degree pan and 90-degree tilt tracking with no additional code, targeting iOS 17+ with Swift 6.3. It requires a physical DockKit-compatible accessory and a real device - the Simulator cannot connect to dock hardware - needs no special entitlements (though camera apps still supply NSCameraUsageDescription), and communicates with paired accessories through the DockKit system daemon. Accessories are discovered by observing DockAccessoryManager.shared.accessoryStateChanges for docked and undocked events. System tracking is the default mode: built-in ML inference detects faces and bodies and drives the motors, toggled with setSystemTrackingEnabled, and users can pick a subject via selectSubject(at:) or selectSubjects. For custom control, apps disable system tracking and supply their own DockAccessory.Observation values - of type .humanFace, .humanBody, or .object, using normalized lower-left-origin rects compatible with Vision - passed with a CameraInformation constructed from the active AVCaptureDevice and current sample buffer at 10-30 fps. Framing is controlled through modes (.automatic as the documented default, plus .center, .left, and .right) and a region of interest set with normalized upper-left coordinates, useful when overlays or non-standard aspect ratios crop the frame. Direct motor control is available after disabling system tracking by calling setAngularVelocity with a Vector3D in radians per second, where x is pitch/tilt, y is yaw/pan, and z is roll. The skill also covers animations, tracking state and subject selection, accessory events, and battery monitoring, with a review checklist for custom-tracking code.