The design of robotic systems for human-hand tasks could benefit for a more intuitive environment for task definition and prototype testing. Virtual reality with human-limb motion identification and visualization seems to be an appropriate environment both for defining the task and to visualize the solution. In this work we present the combination of a method for the kinematic design of robots with serial and tree structures, with a virtual reality and depth-sensing system that allows faithful representation and data extraction for hand and upper extremity motion. The communications and solver are embedded in the virtual reality programming, yielding a tool for design of cooperative human-robot systems.