diff --git a/pages/plural-features/continuous-deployment/resource-application-logic.md b/pages/plural-features/continuous-deployment/resource-application-logic.md
index 624fbe7a..7ad7ede9 100644
--- a/pages/plural-features/continuous-deployment/resource-application-logic.md
+++ b/pages/plural-features/continuous-deployment/resource-application-logic.md
@@ -111,12 +111,17 @@ annotation. Multiple policies can be comma-separated.
 Use `hook-succeeded` and/or `hook-failed` to delete the resource after it completes successfully or fails, respectively.
 If no delete policy is set, the resource will be kept.
 
-{% callout severity="info" %}
-Only jobs and pods are automatically deleted. Other resources will need to be deleted manually or via other automation.
-{% /callout %}
+Only jobs and pods can be automatically deleted. Other resources need to be deleted manually or via other automation.
 
-{% callout severity="warning" %}
 When a manifest of a hook is updated, the operator will reapply the resource even if it has already completed.
+
+{% callout severity="warning" %}
+If a job manifest sets `ttlSecondsAfterFinished`, Kubernetes will delete it after it completes.
+The agent will then see the resource as missing and recreate it unless the manifest also
+includes a hook delete policy via `deployment.plural.sh/sync-hook-delete-policy` or `helm.sh/hook-delete-policy`.
+
+To avoid a delete/recreate loop, ensure the delete policy contains a value we accept: `hook-succeeded` and/or
+`hook-failed`. With a valid delete policy set, completed hooks won’t be recreated unless their manifest changes.
 {% /callout %}
 
 ### Example