diff --git a/CHANGELOG.md b/CHANGELOG.md
index 1578a45d..5ce02cb8 100755
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -17,11 +17,11 @@ Monitor the release status by regions at [AKS-Release-Tracker](https://releases.
 * Bug Fixes 
   * Fix for [LTS](https://learn.microsoft.com/en-us/azure/aks/supported-kubernetes-versions?tabs=azure-cli#long-term-support-lts) to look at AgentPoolResources.
   * PUT managedCluster operations on API versions that didn't support serviceMeshProfile resulted in "invalid mode" error response to the API requests. This issue has now been fixed.
-  * Fix for the wrong MCR URL for Keda in AGC.
+  * A wrong MCR URL for [KEDA](https://learn.microsoft.com/azure/aks/keda-about) image in Air Gapped Cloud was previously used, which may fail the feature enabling This issue has now been fixed.
 
 * Behavioral Change
   * [Artifact Streaming](https://learn.microsoft.com/en-us/azure/aks/artifact-streaming) can now be enabled on an existing nodepool.
-  * The default ResponseHeaderTimeout (which specifies the amount of time to wait for a server's response headers after fully writing the request, does not include the time to read the response body) is now set to 55s to prevent abnormal slow operation caused by hanging connection.
+  * The default ResponseHeaderTimeout (The timeout of http client used by AKS RP, which specifies the amount of time to wait for a server's response headers after fully writing the request, does not include the time to read the response body) is now set to 55s to prevent abnormal slow operation caused by hanging connection.
   * The memory limit for [Azure Key Vault provider for Secrets Store CSI Driver](https://learn.microsoft.com/azure/aks/csi-secrets-store-driver) is now increased from 200 Mi to 300Mi.
   * Expanders flag is removed from [AutoscalerProfile](https://learn.microsoft.com/en-us/azure/aks/hybrid/work-with-autoscaler-profiles#profile-settings) from 2023-11-01-preview since it may cause confusion with existing Expanders. 
   * Config checksum is added to cns daemonset spec will roll CNS pods that need to pick up new config values in azure-cns-configmap, this is to resolve the recent Windows goroutine leak.