Significant attention has recently gone to skeleton-based human action recognition. The goal is to extract features from human skeletons and estimate human poses. However, current methods only capture action information while in real-world applications like cognitive assessment, it is essential to measure executive functioning to help psychiatrists identify mental diseases such as Alzheimer’s, Schizophrenia and Attention-Deficit/Hyperactivity Disorder (ADHD). In this paper, based on the embodied cognition theory, we introduce the Automated Movement Examination (AME) model, a cognitive assessment system using skeleton-based action recognition. AME integrates a pose estimator to extract human body joints, predict the human action and then automatically measures executive functioning. We designed two score functions to measure executive functioning: step score and time score. Our model was evaluated on two different datasets, demonstrating that our approach significantly outperforms existing methods. To make the results reproducible, the code and the details of the experimental setup are made available online at: https://bit.ly/460G9az