Redundant robotic systems are designed to accomplish multiple tasks simultaneously. Tasks are functions of the system configuration, and can be divided into groups by their priority. System redundancy can be exploited by including lower-priority optimization tasks within the control framework. However, it is important that the inclusion of such lower-priority tasks does not have an effect on higher-priority safety-related and operational tasks. This paper presents a novel task-priority framework based on a hierarchy of control Lyapunov function (CLF) and control barrier function (CBF) based quadratic programs (QPs). The proposed method guarantees strict priority among different groups of tasks such as safety-related, operational and optimization tasks. Moreover, a soft priority measure in the form of penalty parameters can be employed to prioritize tasks at the same priority level. As opposed to kinematic control schemes, the proposed framework is a holistic approach to control of redundant robotic systems, which solves the redundancy resolution, dynamic control and control allocation problems simultaneously. Simulation results of a hyper-redundant articulated intervention autonomous underwater vehicle (AIAUV) is presented to validate the proposed framework.