Tensegrity robotic joints, inspired by the musculoskeletal systems of vertebrate animals, have gained interest due to their unique interplay of tension and compression forces, offering a high strength-to-weight ratio and inherent adaptability. However, their use in legged and grasping applications remains challenging. A key challenge is striking the right balance between high compliance, which can undermine stability and control, and high stiffness, which can restrict essential movement and demand more powerful actuators. This study presents a tensegrity-inspired spine-like joint that integrates thin McKibben pneumatic artificial muscles and rubber cords directly into its tensile network. The artificial muscles enable active joint bending while also serving as tensile elements. The rubber cords counteract these forces, storing elastic energy to facilitate smooth recovery to the original position. The joint's topology design is primarily influenced by the artificial muscles’ limited 20% contraction. To optimize the balance between range of motion, payload capacity, and structural stability, the joint's geometry is refined through an evolutionary algorithmic form-finding process. Two generated joints are integrated into two functional robotic applications: a crawling robot and a gripper, showcasing their adaptability for diverse robotic applications.