The rapidly increasing quantity and diversity of data stored on our PCs made locating information in this environment very difficult. Consequently, recent research has focussed on building semantically enhanced systems for either organizing or searching data on the desktop. Building on previous work, in this paper we present the Beagle++ toolbox, a set of extendable building blocks for implementing such a system. The corresponding modular desktop search architecture integrates our previously developed metadata generators and ranking components, uses an RDF database to share data between components, and can easily integrate other external components to improve desktop search quality. Additionally, we provide implementation details about all our current components, how they interact with each other, and how to install the complete system on top of a Linux distribution.