Accepting a build only from a repository's content-CI workflow has one way to go wrong: the platform looks for that workflow where the repository does not keep it. Every delivery is then refused, GitHub is answered normally, and every Space syncing that repository quietly stops receiving content — the failure that once left a Space 38 hours behind with nothing reporting a problem.

Two changes close that. Systemorph/Memex, whose deployment records two portals sync, has no ci.yml and is now declared by the platform, so it keeps publishing. And when a repository that a Space actually syncs has a green run refused while no run of its expected content CI has ever been accepted, the portal logs a warning naming the workflow it found, the one it expected, the Spaces that are frozen, and how to declare the right one — instead of failing silently.