In general I like the idea of the proposal and think it will make the interoperability testing easier to use in helping me and others test and debug interoperability issues. I did have one point that I would like more clarification on in the proposal.
My question really is about who can create tests and who can call the hosted endpoints. I am envisioning a scenario that does not seem to be addressed in the proposal. The scenario I would like to see is that a user can run the tests or see the results of previous tests between hosted endpoints that have been set up by the Stonehenge contributors (steps 8-10 of the walkthrough). They could then call any of the endpoints with their code and if the call fails examine the messages to see what is different.
When I look at the HostedClientResult class it has 2 properties (requestMessageLog and responseMessageLog) that would appear to allow me to retrieve the entire contents of the messages sent across the wire. If this is the case I can see where it would be very valuable to have the messages captured from a successful test in the documentation for that test so I could compare that to my failed case.
Along with that I wasn't sure what was meant by the statement "Results are stored." in step 10 of the walkthrough. If there are multiple results stored (including any that I run) then searching through a large set of results might actually deter me and other users from trying this. I would propose that there be 1 result stored long term as the "correct" results from the test. This could even be added to the documentation for that scenario. Users could then call any of the end points and retrieve the messages from the results but these would not be stored by apache.org.