Max/MSP/Jitter will certainly easily allow this type of tool to be realised, but apart from someone actually building a complete patch that does all this for you, there really is only one way to achieve this: by building and learning; break your interaction/feedback environment down into small manageable parts, and attempt to implement each one, slowly adding more complex input/output as you learn. If the Lemur is outputting midi or OSC data, you need to look at [midiin] or [ctlin] or [udpreceive]; for soundfile triggering you should look at [play~] [buffer~] and [sfplay~]; as for video control, Jitter is a closed book to me, but post on the Jitter forum and you might get help there.