In this paper, we present the design and development of a game-assisted stroke rehabilitation system RehabFork that allows a user to train their upper-limb to perform certain functions related to the task of eating. The task of eating is divided into several components: (i) grasping the eating utensils such as a fork and knife; (ii) lifting the eating utensils; (iii) using the eating utensils to cut a piece of food; (iv) transferring the food to the mouth; and (v) chewing the food. The RehabFork supports the user through sub-tasks (i)-(iii). The hardware components of RehabFork consist of an instrumented fork and knife, and a 3D printed pressure pad, that measure and communicate information on user performance to a gaming environment to render an integrated rehabilitation system. The gaming environment consists of an interactive game that utilizes sensory data as well as user information about the severity of their disability and current level of progress to adjust the difficulty levels of the game to maintain user motivation. Information pertaining to the user, including performance data, is stored and can be shared with care providers for ongoing oversight.