Rehabilitation therapy based on motor imagery brain-computer interfaces (MI-BCIs) has been shown to produce lasting improvements in upper limb motor functions after stroke. However, a significant percentage of novice users (10% to 30%) fail to adequately control them, partly because traditional stimulation protocols are neither enganing nor transparent, hindering learning. Furthermore, these protocols do not provide meaningful real-time feedback that allows the user to efficiently adjust the brain activity modulation strategy employed. Here we present a videogame-based stimulation protocol for MI-BCIs with applications for upper limb rehabilitation that integrates supportive backward adaptation (SBA) to provide informative online feedback reflecting how well the instructed MI task was performed. Using this protocol, to date a database comprising five sessions from four healthy users (three males and one female, 21-33 years old, right-handed) has been acquired. This database was recorded using affordable, portable, non-clinical-grade acquisition systems. Preliminary results indicate that higher task accuracy was associated with lower median and deviation values of the SBA support index, suggesting that this metric can serve as an indicator of user progress across sessions. Participants reported being highly engaged in the task, considered the feedback clear and useful, and the game motivating.