With the rise of automation comes the need for automated decision-making. Real-world domains of business and industry often utilize automated agents, like robots, to perform various tasks. Automated agents rely on decision-making software to decide on a course of action to complete such tasks. When decision-making software chooses a sequence of actions for an automated agent to perform, a plan is created. This process of planning as well as decision-making software are subjects of study in the research field of Automated Planning, where a decision-making software is referred to as AI Planner. The manner in which an AI Planner conducts planning depends on the planning technique employed. One of the more widely used planning techniques is Hierarchical Task Network (HTN) planning, which composes available actions of a domain into a hierarchy, to allow for planning, resembling the way a human individual would conduct planning. A crucial factor in human planning is the notion of risk, since, in real-world domains, actions frequently lead to one of many possible results. Often there is no guarantee for a particular result, and some results might be undesirable. Therefore, when an individual performs planning, they take risk into account, as the individual stands to lose a certain amount of resource, such as money or time. The way an individual approaches planning with risk in mind, shows an attitude towards risk. An individual, avoiding risk bearing actions, shows a risk averse attitude, while an individual, taking risks, shows a risk seeking attitude. If no risk consideration is made, the individual shows a risk neutral attitude. For particular real-world domains, a reasonable individual would show a specific risk attitude out of necessity. Consider aviation, for example, where many would agree, that risk avoiding actions should be chosen. When risk bearing actions are chosen in such high stakes domains, undesirable results can include the loss of human lives. Therefore, dealing with risk by showing the appropriate risk attitude, is especially important in such domains. Since automated agents are often employed in those domains, AI Planners should be aware of risk and factor risk into planning. Furthermore, AI Planners should be able to perform planning according to an appropriate risk attitude, so automated agents can act according to this risk attitude. However, as far as we know, a risk-aware AI Planner is a rarity in the research field of Automated Planning. To contribute towards risk-aware planning, we present a proof of concept for an approach to risk-aware HTN planning. The approach employs utility functions to heuristically guide planning towards plans, that adhere to a specified risk attitude. The proof of concept comes in the form of Risk Aware PANDA3 (RAPANDA3), an extension to the AI Planner: Planning and Acting in a Network Decomposition Architecture 3 (PANDA3). On top of that, we propose an addition to an HTN input language, which allows the modelling of risk for actions. Furthermore, we provide multiple self designed HTN domains, which use this addition. With these domains, we conduct an evaluation of RAPANDA3, to scrutinize the implemented approach to risk-aware HTN planning, and to show, that this approach results in plans, adhering to a specified risk attitude.