We present a Model Driven Engineering (MDE) approach to dependability evaluation of component-based coherent dyadic systems, implemented by the FaultFlow library, combining simple high-level modeling with powerful quantitative evaluation methods. In the functional perspective, distinctive features are: modeling of fault propagations within individual components and between different components, possibly not connected through physical or communication interfaces; support for non-Markovian distributions, both for the times to the occurrence of faults and for the duration of fault-to-failure propagations; derivation of the distribution of the time to the occurrence of a given failure; derivation of fault importance measures, for models where each fault does not propagate into multiple failures and, viceversa, each failure does not act as fault to multiple components, achieving evaluation efficiency even for significantly complex systems with hundreds of different faults. In the implementation perspective, distinctive features are: definition of a custom-made extensible metamodel to specify the system structure and failure logic; automated derivation of metamodel instances from Systems Modeling Language (SysML) Block Definition Diagrams (BDDs) and Stochastic Static Fault Trees (SSFTs); automated derivation of the mentioned dependability measures; open source availability. We illustrate the typical modeling and evaluation workflow with relevant uses cases, comparing functionalities with those of other dependability evaluation tool.