Yes, you can access blobs too. I've been using the HTTP API (the one that backs CloudKitJS) to process data in app engine. So far I just used the JS API token for authentication. Records with an associated blob asset provide a temporary accessible URL to the asset which is usually stored on S3.
Not sure about latency of the requests, never had a problem with that. More often the problem was that I had confirmation my client software had written data to CloudKit but it might take a while before the server can actually query for it. (Could be an indexing delay haven't checked if the object is immediately available using the object ID)
What will happen to it when they will rename to 'Mozilla Personas'? Will the login screen suddenly show the Mozilla Persona logo instead of BrowserID as it currently does? I can't imagine this gaining any traction with bigger sites / brands if it is pushing the Mozilla brand.
While I do like the thought of using CoreData and code I've written for other apps on the web, the following sentence immediately made this thing much less interesting:
"WebAppKit runs exclusively in garbage collection mode to minimize the memory management burden."
I've never liked garbage collection, if you find memory management a burden, you're doing it wrong. And now that Apple introduced ARC even Apple recommends against using the GC.
Not sure about latency of the requests, never had a problem with that. More often the problem was that I had confirmation my client software had written data to CloudKit but it might take a while before the server can actually query for it. (Could be an indexing delay haven't checked if the object is immediately available using the object ID)