Among the several categories of trust models, cognitive models have important features. Initially these models were only informally defined, but formalizations were recently proposed. The concepts of the models are thus sufficiently well defined to be implemented and evaluated. In this paper, the cognitive trust model proposed by Castelfranchi and Falcone is integrated into a BDI (belief, desire, intention) agent architecture and implemented with the Jason programming language. The ART testbed scenario is then used to experiment and evaluate both the model and the implementation.