I can't go through all the steps in a lot of detail here but I can give a basic overview.
I already had a development environment set up for Google Appengine on my computer. This uses PyDev in Eclipse. I'm not much of a Python programmer but I've done a little bit for Appengine, mostly just following the Google tutorials. For my robot, I created a new Google Appengine project and I created a new project hosted on Google Code (using Subversion because I'm comfortable with it). You can see the code for Q&A bot in my qnabot project.
Then I checked out the source code for that project to my computer. I created a project in Eclipse pointing at the new working copy of the source. I grabbed the Wave robot Python client and had a look through it. I copied the source of the dummy sample, which currently just has an app.yaml and a very short listing for the robot source:
Since I'm not really a Python programmer it took me a while to realize that the api package mentioned in the source was the contents of the webapi directory in the same download. I just copied the webapi directory into my source directory and changed the lines in the dummy sample source to reference "webapi" instead of "api." I deployed to appengine then went over to the Google Wave sandbox.You have to have an invite to get in right now, I'm if you want to add me. I got my account because I was at Google I/O this year.
In Google Wave, I added as a contact. Then I click on the new contact and the profile shows up, including a button to create a new Wave with qnabot as a participant. When I clicked to create a new wave, the robot added its "Hello there!" message to the first blip.
I did have a couple false starts (though the whole thing took less than an hour). One helpful tool was the view of my appengine logs. I could see what URLs were being requested and what errors were occurring. That's how I knew the api package was the webapi stuff and I could look at the /_wave/capabilities.xml url as an example of something that should show up on my local development environment. It also helped to read the tutorial for the Java Wave robot client library which is already out there.