Clustered tensegrity mechanisms have elicited extensive attention in recent research due to their easy control system and high stiffness-to-mass ratio. However, modeling and analyzing these mechanisms are still challenging due to the clustering of cables and redundant structural parameters. This article proposes an energy-based kinematic modeling method for a modular clustered tensegrity mobile robot. The design of the clustered tensegrity robot is inspired by the biomechanics of worms, allowing it to achieve two locomotion modes resembling earthworm-like and inchworm-like movements using two motors. Moreover, the clustered and modular structure enables the robot to increase the number of modules as needed without increasing the number of actuators. This feature enhances the robot's terrain adaptability without adding complexity to the control system. The article establishes kinematic models using the energy method and clarifies the motion law of nodes on the sliding cables of the robot, considering multiple structural parameters for both locomotion modes. Based on these models, the article reveals the mapping relationships among various structural parameters (i.e., cable-hole gap, cable-hole friction, stiffness and original length of elastic cables, and ground–robot friction) and locomotion performance (i.e., morphology, displacement, and velocity) of the robot. Furthermore, structural parameter optimization is performed to enhance the kinematic performance of the robot in both locomotion modes simultaneously. To validate the proposed kinematic modeling method, a prototype with two modules is developed, and experiments are conducted to assess the robot's locomotion performance. These experiments demonstrate the effectiveness and rationality of the proposed method.