A modelling procedure consists of a sequence of modelling tasks. With the increasing size of models during the development, the relations among modelling objects and their corresponding tasks become more and more complex. Therefore, the automatic identification of available tasks for current modelling environments can significantly improve the efficiency of modelling. In this paper, we present a novel approach which integrates task pre- and post-conditions into ontologies that describe the models, and infers availability of tasks through ontology reasoning and query answering. We illustrate this approach with an application on process modelling and show how this can be generalised to other domains.