Programming of robots based on the sequencing and execution of high-level function blocks (advanced skills) is a method of increasing flexibility, which is necessary to enable applications such as the dismantling of old appliances or waste from electrical and electronic equipment (WEEE). Such skills are characterized by their capability of acting autonomously depending on context and situation. This requires the management of model data. Here, we show an application of the Asset Administration Shell (AAS) for making data on static robot models as well as dynamic state available to domain-specific software components. The system shows the use of such digital twin for programming and action generation of intelligent robots. We used a bottom-up approach that relies on the integration of already available data representations while adding information necessary for practical use.