Declarative pattern mining implies to define common frameworks and atomic operators for different problems. In this paper, we consider INclusion Dependency (IND) mining which is a classical data mining problem, with many applications in databases and data analysis. We present a novel and quite surprising result: IND mining can be optimized by a closure operator, as it is done for support-based pattern mining. As a consequence, and through a data pre-processing, satisfied closed INDs can be mined with very few programming efforts, using closed itemset mining procedure as a basic operator.