Currently our appstream data is attached to aptly via a Jenkins job that runs after snapshotting. Since we only snapshot user/ this ultimately means that one has no appstream data unless user/ is enabled. It also means that re-publishing a snapshot outside of jenkins (i.e. via api) would break appstream data.
Also the appstream generation is highly fragile as it involves a lot of untested code.
The entire shebang should be built into aptly