Question 440
A production Azure security implementation shows this failure: A security engineer deploys a new Azure VM with no endpoint agent and assumes the VM cannot be assessed for known software vulnerabilities. Which recurring validation best protects against the observed problem?
Explanation
Correct Answer
This guardrail addresses defender for servers agentless vulnerability scanning: Supported Defender for Cloud agentless machine scanning can assess certain VM properties and installed software without relying on a traditional guest agent. The prevention measure is: Review supported scan coverage and inventory accuracy when adding agentless-protected compute resources. A control in another subsystem would not prevent this specific recurrence.
Incorrect Answers
Inventory visibility and generic baseline recommendations do not replace the dedicated supported Defender for App Service plan. “Include App Service protection in onboarding checks for Internet-exposed web applications” would address that alternative cause, not the condition in this question: supported Defender for Cloud agentless machine scanning can assess certain VM properties and installed software without relying on a traditional guest agent.
“Continuously combine vulnerability severity with external reachability and asset importance in remediation planning” responds to this separate finding: patch priority should consider vulnerability severity, exploitability, exposure and business criticality instead of a simple ticket age queue. This question instead needs “Review supported scan coverage and inventory accuracy when adding agentless-protected compute resources”, so the alternative does not meet the stated condition.
Control-plane Resource Manager activity can require dedicated detection separate from guest server security events. “Keep management-plane threat detection enabled for governed subscriptions independent of server-agent deployment” applies to that separate problem. The observed case is a security engineer deploys a new Azure VM with no endpoint agent and assumes the VM cannot be assessed for known software vulnerabilities; choosing this alternative would address the wrong failure mode.