This version of the WS-Agreement specification is much clearer than previous versions. In my opinion, it is ready to 
become an OGF proposed recommendation. This will greatly facilitate further development of the implementations that 
already exist. 

I notice a couple of minor errors or inconsistencies (see below). I think that these, along with other points raised 
during public comment, can be addressed in a minor revision, which would not require yet another public comment period 
before being published as an OGF proposed recommendation. 

1) In section 7.1 (Agreement States), the text
"Terminated - The terminated state means that an Agreement offer has been terminated by the Agreement Initiator and that
 the obligation no longer exists. This state MAY follow Pending, PendingAndTerminating, Observed or 
ObservedAndTerminating when the termination decision is made." implies that a transition from Pending directly to 
Terminating is possible, yet no such transition is depicted in the diagram of the agreement state model.

2) In the same section, the diagram of the agreement state model shows
transitions from "PendingAndTerminating" to "Pending", and "ObservedAndTerminating" to "Observed". While I can 
understand the forward transitions (e.g. Pending to PendingAndTerminating), the reverse transions are puzzling, and not 
explicitly mentioned in the text. The text does say:
"There may be domain dependent cases where an agreement completes normally while in PendingAndTerminating or 
ObservedAndTerminating states. These cases can be handled by allowing a domain-dependent sub-state of Terminated to 
indicate a normal completion prior to termination completion."
But this doesn't quite explain the transitions in question.

3) Section 8.2 Forms of Acceptance.
The second last sentence:
"The initiator MAY still determine the result of by other means such
as querying the status of the resulting Agreement."
appears to be missing some words between "of" and "by".