Upper-limb motor impairments are among the most disabling consequences of neurological conditions, severely affecting activities of daily living (ADLs) and creating a persistent need for rehabilitation and assistive interventions. Functional arm movements are essential for restoring personal independence, yet restoring natural arm–hand coordination remains a major challenge in rehabilitation robotics. Full-arm hybrid exoskeleton–functional electrical stimulation (FES) systems represent a promising solution, combining the mechanical support of rigid exoskeletons with FES-induced neuromuscular activation. Although proximal-distal coordination during reaching is well established in motor control literature, its use as predictive information for hand-state inference in hybrid control remains largely unexplored. Synergistic control approaches address this by inferring hand movement intentions from residual proximal arm functionality, but existing methods are generally limited to simple reaching tasks and rely on phase- or time-based segmentation, restricting applicability in unstructured real-world scenarios.