I hate helm. For us, it led to a PITA knowing where values came from. You had to debug logic in the helm charts to know why a value existed in the cluster with no good debugging tools.

We switched to kustomize with tooling around it. It is dead simple to know why a value landed in an environment this way at the expense of yaml proliferation but I would rather have tooling to help us manage that than trying to debug why a value got into production due to an bug in helm logic in the templates. I don’t want to debug service logic and helm templating logic.