Current PDDL tool use plans for robotic manipulators are made actionable by supplying extra semantic information or structuring actions into phases. However, these solutions are not compatible with cycle-based affordances such as pounding or wrenching and do not directly link actions to evaluations for closed-loop control or learning methods. In this work, the cognitive linguistics theory of image schema is used to structure affordance-based tool use motion planning and tool use evaluation within spatiotemporal elements. The resulting method semantically breaks down an affordance's grasp, align and cycle or contain steps into specific image schemas and their literature defined elements. These elements are associated with actions from a robotic controller or sensors. Each action has an associated feature to evaluate such as checking if the subject has reached its goal pose after each cycle. The result of this feature is fed back to the motion planner to decide if the action cycle should be repeated or not. Semantic features found from associated actions are given context by categorizing them as task, subject, tool, or environment features. It is found that these methods help structure the process of tool use and enable cycle-based tool use tasks as proven through a simulation experiment with the pound affordance. Image schemas semantically break down tool use tasks by their specific spatiotemporal elements, providing an avenue for robotic agents to understand and learn tool use more deeply than existing methods.