Okay, received in the mail (finally). The Seeedstudio's source is a company called JYE Tech,
http://www.jyetech.com/ - thankfully with an English page as well.
Although this little oscope is no speed demon, it is surprisingly capable. When you unbox the unit, it's even smaller than the pictures suggest. The whole thing is the size of two Arduino Duemilanove boards side-by-side. I got the preassembled form, which comes with a pcb faceplate and backplate to protect the internals better. This is probably a good thing, since I noticed at least one hand-patched revision to the oscope's motherboard, and the kit form probably does not include written errata.
The probes are needle-tipped alligator clips, plugged in with a low-tech RCA jack. If left floating, you can see the signal stray by a few millivolts. If you touch both probes, you get a nice wavy 2Vpp biological signal. The unit has a tiny jumper wire you can clip that provides a test signal: a +5V square wave at 500Hz. They suggest you learn the functions with this test signal.
You can adjust the volts/div scale, the sec/div scale, and raise or lower the location of 0V in the view. This is a triggering storage scope, so there are three modes to display: free running (AUTO), refresh on every trigger (NORM), and freeze the display to show the first trigger (SING). The level required to trigger, and the direction of change required to trigger, are easy to adjust. For example, you can trigger on any downward change through +3V. The scope will show 256 samples in the vicinity of that trigger point. You can shift the view to focus on more of the time before or after the trigger moment.
The only real downside is that the display can only update at about 2~3 frames per second. I'm sure it's due to the LCD driver that they chose. If you hit OK, the screen will HOLD the current contents of the sample buffer (which is always newer than what you're seeing). If you're in NORM mode, this means you can freeze the screen to study a packet of data as it goes by, but due to the update lag you might expect you can see a signal of interest and hit OK to freeze it. You can't.
Oh, and a pleasant surprise: it seems to store your selection preferences in nonvolatile memory, so it starts up again with the same time and trigger settings you used last time.
With these features, it was pretty easy to study the i2c conversation between a Wii nunchuck and the Arduino. From the source code, I knew that the first byte of all command packets was a constant 0x52 so this should be an easy pattern of 1s and 0s to spot on the line.

There's also a frequency-counting mode which I have yet to try. It was hard enough tearing myself away from the fun i2c study that I was doing to take a photograph, but now I'm off to see what a PWM-to-audio waveform looks like.