Community deception is about concealing the members of a community $\mathscr {C}$ within a community structure (set of communities) discovered by a community detection algorithm. This issue has only been studied from an edge-centric standpoint. This article introduces nDec, a node-centric community deception approach that can directly influence where $\mathscr {C}$’s members are placed. Indeed, nDec considers how nodes can be moved between communities and analyzes the most profitable $\mathscr {C}$’s node removals and additions. nDec is built around a node-centric view optimization function based on modularity, a well-known cluster quality metric. We discovered that rather than blindly minimizing modularity as edge-centric approaches do, it is possible to devise a strategy appealing to detection algorithms while improving the level of concealment of $\mathscr {C}$. Based on a thorough evaluation and comparison with the state-of-the-art, we conclude that node-centric deception provides a more comprehensive and flexible strategy than the edge-centric state-of-the-art.