Background Cognitive and motor impairments are major determinants of functional independence after stroke. While robot-assisted training is effective for motor recovery, its impact on cognitive function through integrated cognitive–motor training remains less explored. This study evaluated the effects of an end-effector robot-assisted integrated cognitive–motor rehabilitation program on global cognitive function and upper limb motor outcomes in patients with chronic stroke. Methods In this single-center, randomized, assessor-blinded exploratory trial, 30 patients with stroke were randomized into an intervention group ( n = 15) or a control group ( n = 15). The intervention group received robot-assisted cognitive–motor training using the Rebless Planar system (30 min/session, 10 sessions over 3 weeks), which integrated goal-directed reaching with tasks involving attention, memory, and executive function. The control group received time- and dose-matched conventional paper-based cognitive rehabilitation. Primary outcomes were the Korean Mini-Mental State Examination (MMSE) and the Korean Montreal Cognitive Assessment (MoCA). Secondary outcomes included CERAD subdomains, Fugl–Meyer Assessment of the Upper Extremity (FMA-UE), and the Modified Barthel Index (MBI). This study was registered with the Korean Clinical Research Information Service (KCT0010564). Results Both groups exhibited significant within-group improvements in MMSE and MoCA, with no statistically significant between-group differences. For secondary CERAD subdomains, observed within-group gains did not maintain statistically significant between-group differences after False Discovery Rate adjustment. However, upper extremity motor recovery was significantly superior in the intervention group (FMA-UE 33.6 to 37.5) compared to controls (25.7 to 25.9) ( p = 0.012, Cohen’s d = 0.98). Conclusion Robot-assisted integrated cognitive–motor rehabilitation using an end-effector system is feasible and safe in chronic stroke patients. While both robotic and conventional cognitive rehabilitation led to comparable improvements in global cognition, the robot-assisted group demonstrated superior upper limb motor recovery. These findings suggest that integrated dual-task training is a promising strategy for post-stroke motor recovery, though its incremental cognitive benefit over conventional therapy requires further investigation in larger trials. Clinical trail registration This study was registered with the Korean Clinical Research Information Service (KCT0010564).