I agree the implementation isn't sound, but there's nothing wrong with the concept of a shaft passer.

As an example of why it is conceptually feasible, remember that you could computationally (or mechanically) determine when a shaft is closing in and open a gap in that particular support at the right time, maintaining ability to pass torque through all other supports.