Attempting to clear up some issues, clicked on the Clear Cluster Cache button as it was implicated in fixing Helm errors some time ago.
This message makes the operation feel not overly unsafe:
Clear the local cache of namespace and service accounts. This is necessary if your integration has become out of sync. The cache is repopulated during the next CI job that requires namespace and service accounts.
When I re-run my AutoDevOps CI/CD pipeline however :
 ++ auto-deploy ensure_namespace  error: the server doesn't have a resource type "namespace"  error: You must be logged in to the server (Unauthorized)
The various KUBE* variables seem to be set, so I am not sure why its failing to auth.