Researchers have introduced GOAG (Generative and Object-Agnostic Grasp Planner), a novel deep generative model designed to improve dexterous robotic manipulation by overcoming the limitations of object-specific training. Unlike traditional deep-learning planners that struggle to generalize to novel geometries, GOAG utilizes the principle that gripper and object surfaces share identical geometry at contact points. The method works by learning a compact latent representation of a specific gripper's contact surface distribution. By introducing object features only during the inference stage, the model can efficiently sample valid, admissible grasp configurations that are compatible with the gripper's physical capabilities without requiring object-specific datasets during training. Experimental validation conducted in both simulated and real-world environments demonstrates that GOAG achieves state-of-the-art performance on the MultiDex dataset, reaching an average success rate of 86.93%. Furthermore, the approach offers significantly faster processing speeds when generating large numbers of potential grasps compared to existing methods. By decoupling the learning of gripper capabilities from specific object geometries, GOAG provides a scalable, object-agnostic solution that effectively addresses the generalization challenges prevalent in current data-driven grasping frameworks.