Upper-limb motor impairment severely limits personal autonomy, yet access to rehabilitation is often restricted by heavy clinical workloads and high equipment costs. (1) Traditional devices frequently rely on virtual reality environments that induce dizziness or motion sickness. (2) This research proposes an accessible, lightweight robotic therapy platform. The system optimizes clinical operational capacity by allowing a specialist to oversee structured routines via a centralized network, combining interactive social robotics (NAO humanoid robot), named a master with mechanical motor assistance (slave) to improve patient adherence. The mechatronic architecture of the robot features a central controlled by Python, that server running with a graphical interface (Gradio) and a database (SQLite) that will allow to manage and log therapeutic routines such as extension, and cylindrical grasp. Using a dedicated Wi-Fi network via the UDP protocol, the server simultaneously transmits synchronized angular coordinates to two distinct endpoints: a humanoid robot—acting as a visual and motivational guide—and a wearable three-finger hand exoskeleton, constructed from flexible Thermoplastic Polyurethane (TPU). The wearable system is actuated by a servomotor using tendon-like cables, and controlled by an ESP32 microcontroller to ensure real-time processing. Initial execution scripts showed that, following a 10-second connection setup, synchronization commands operate reliably. The servomotor responds in approximately 1.0 second. Although this initial latency exceeds the strict 200 ms threshold due to the robot's internal processing times, the logical flow remains stable and secure. In summary, this study successfully implemented a centralized robotic architecture that integrates a 3D-printed ergonomic chassis with social robotics.