We had to learn that no current toolkit approach available from research or commercial environments[MABY96] could provide us with the functionality needed to meet the use cases of application control in a large distributed data acquisition system. Although each product fits a part of the requirement space, no tool exhibits enough flexibility that allows all aspects to be covered. The fundamental problem is a lack of adaptability in the available solutions. A framework that avoids any freezing of hardcoded policies can represent a remedy to this situation. By encapsulating user applications as well as core system functions into a single narrow interface, a fully configurable system can be achieved. According to the evolving needs of the users, new policy objects can be provided. As the subject of control are the task capsules and not the applications themselves, any program can be included in the system, as long as it is possible to program a command mediator. In order to demonstrate these facilities, we designed and implemented a proof of concept system making use of standard software design patterns and a platform independent communication subsystem. Although COTS tools point in exactly this direction, we see that platform dependence is still a major issue. Even if systems provide the possibility to supply alternative policy objects, the user is bound to support from the toolkit vendor. To our best knowledge no design as it has been presented here has been proposed elsewhere.