apiVersion: helm-expt.confighub.com/v1alpha1
kind: ProductionDispositionReceipt
metadata:
  name: prometheus-community-kube-prometheus-stack-crd-lifecycle-and-upgrade-policy
spec:
  chart: prometheus-community/kube-prometheus-stack
  version: "85.3.3"
  disposition: CRD lifecycle and upgrade policy
  decision: accepted
  acceptedAt: "2026-06-09"
  scope:
    - local-test
    - production-review-input
  summary: >-
    The default kube-prometheus-stack base renders Prometheus Operator CRDs as
    reviewed objects. The no-crds base omits them for clusters where the CRDs
    are managed separately. This disposition accepts both CRD ownership modes
    as production review input. It does not choose the production CRD owner or
    claim CRD upgrade safety.
  evidence:
    - path: recipes/prometheus-community/kube-prometheus-stack/85.3.3/control-points.yaml
      claim: The recipe records CRD policy as variant-controlled with 10 CRDs in default and none in no-crds.
    - path: recipes/prometheus-community/kube-prometheus-stack/85.3.3/value-model.yaml
      claim: The value model records crds.enabled and CRD selection controls explicitly.
    - path: recipes/prometheus-community/kube-prometheus-stack/85.3.3/revisions/default/r001/rendered/object-inventory.yaml
      claim: The default rendered object inventory records the Prometheus Operator CRDs.
    - path: recipes/prometheus-community/kube-prometheus-stack/85.3.3/revisions/no-crds/r001/rendered/object-inventory.yaml
      claim: The no-crds rendered object inventory omits CRDs while retaining Prometheus Operator custom resources.
    - path: recipes/prometheus-community/kube-prometheus-stack/85.3.3/revisions/default/r001/receipts/scan-receipt.yaml
      claim: The default rendered-object scan records CRD upgrade policy warnings for the CRDs.
    - path: recipes/prometheus-community/kube-prometheus-stack/85.3.3/revisions/default/r001/receipts/helm-equivalence-receipt.yaml
      claim: The default base is Helm-equivalent under recorded inputs.
    - path: recipes/prometheus-community/kube-prometheus-stack/85.3.3/revisions/no-crds/r001/receipts/helm-equivalence-receipt.yaml
      claim: The no-crds base is Helm-equivalent under recorded inputs.
    - path: runs/top20-local-kind/kube-prometheus-stack-default/observation-receipt.json
      claim: The local-kind observation records CRD bootstrap apply and Established checks for the default base.
    - path: runs/live-kind-parity/prometheus-community-kube-prometheus-stack-no-crds/receipt.yaml
      claim: The no-crds live parity receipt blocks because Prometheus Operator CRDs are missing from the target cluster.
  affectedVariants:
    - default
    - no-crds
  acceptedPolicy:
    default: Accepted as the first production-review base when this package owns the Prometheus Operator CRDs.
    noCrds: Accepted as review input only for targets with compatible Prometheus Operator CRDs already staged and observed.
    upgrade: Chart upgrades must review Prometheus Operator CRD schema compatibility before replacing the supported version.
  variantCaveats:
    - "no-crds is not standalone; it requires compatible CRDs in the target cluster before apply."
    - "CRD render parity does not prove Prometheus Operator upgrade safety."
  remainingProductionBlockers: []
  nextDecision: A production support decision should choose CRD ownership, CRD upgrade ordering, compatibility checks, and rollback procedure for the target scope.