Well, the setup was easier than I expected- after checking the pins on my scope to verify the data pin, I realized I could just use the Prop Plug as a 3.3V serial-to-USB interface to my PC. I soldered some pins onto the iSobot servo comm pin (adjacent to GND) for easier clip connecting, and used the ground pin of the 5-pin header in the middle of iSobots controller board. Here's an image of the setup for recording the left leg- only 2 wires needed!

Next I configured Hyperterminal for 2400 baud, 8-N-1 with no flow control. I started a 'capture to text file' session and then turned on iSobot. Voila! Well, not quite perfect, but good enough to make me happy. I had to open the saved file with UltraEdit to see the hex values, but there it is- strings of 8 bytes, starting xFF x05... In a few cases a byte was dropped, and the very first string after power up was recorded as xE8 x05 x00 x00 x00 x00 x00 x05. I don't know right now if xE8 is an initialization code, or just a recieive error; I'll check to see if that always occurs on powerup, or is variable.

During the startup sequence, iSobot does a little bow and then stands straight up, so not much leg action going on. I'll try using the IR controller to send 'left kick' and other leg commands to see if the numbers coming out can be correlated to servo positions. I believe the x00 sets the servos to 'limp' eg no holding torque. Again that's to be verified. Also to be done is identify which byte position corresponds to each servo.  For now, suffice to say that standing straight up is managed by repeatedly sending the string

xFF x05 x7D x65 xB5 x9A x7E xB6, where the hip and ankle tilt servos (first and last in the chain) are near their center positions. x7E is the center point between x01 and xFF (decimal 126 in a 1-255 range). So I expect it'll be in the same order as the physical servos, either going down the leg or up.