Computer games have emerged in the past decade as potential media beyond entertainment. Despite its popularity, game accessibility remains a major concern of various researchers. Children population with motor disabilities is a potential target for developing entertainment or therapeutic support games due to their interest to play. This paper presents: (1) a framework for mobile games for children with motor disability using simple hand postures and (2) xgboost decision tree as a hand posture recognizer (98.48 percent training accuracy and 96.76 percent testing accuracy) as a prototype of hand posture-based commands as assistive technology to interact with games.