Our organization has recently licensed Silk for testing our Java interfaces and we are trying to push through the learning curve as quickly as possible. Most of our problems to date have centered around insuring that our custom (Swing-derived) objects are recorded in a way that works consistently at run-time and promises some resilience as the software evolves. One piece of knowledge we are clearly lacking concerns how the agent performs its tasks with respect to recording. The only documentation blurb I have found regarding the agent is that it 'translates the commands in your 4Test scripts into GUI-specific commands.' As near as we can tell, the agent also plays a central role in exposing our application for 4Test script recording. How this works does not appear to be addressed anywhere. If anyone has any information on this topic, we would be extremely grateful.