This paper presents a scalable Watchdog Agent®-based toolbox approach for machine health prognostics. The toolbox consists of modularized embedded algorithms for signal processing and feature extraction, performance assessment, diagnostics and prognostics, which can be reconfigured for different machinery prognostic applications, and can be extensible and adaptable to most real-world machine situations. A decision making technique, Quality Function Deployment (QFD)-based tool selection method, is applied for the automatic selection of algorithms from the Watchdog Agent® toolbox using multiple criteria. In addition, the architecture for the Watchdog Agent®-based real-time remote machinery prognostics and health management, which incorporates remote and embedded predictive maintenance technologies, is presented. An industrial case involving the automatic tool changer of a machine tool is presented to illustrate how the Watchdog Agent® toolbox can be used in diverse scenarios.