A database consists of an object space that is information about objects in (some part of) the real world. Every object existing in the real world has properties, including an identity that uniquely distinguishes it from other objects. In a belief database multiple subjects are hypothesized. Subjects have varying beliefs about existence, identities, and other properties of objects. A multilevel security database is a belief database where the subjects form a hierarchy. The upper users can see the object space of lower users but the lower users cannot. In fact lower users may not even be aware of upper users. This paper presents a skeleton based model for storage and query of a multilevel security database. This model follows the parametric approach proposed by Gadia. As in all works in the parametric approach, the central focus of this framework is to support most natural query of data.