Cloud Security Assessment Checklist: 12 Checks Mapped to Microsoft Defender for Cloud - TrustedTech

Cloud Security Assessment Checklist: 12 Checks Mapped to Microsoft Defender for Cloud

Need Help Figuring Out the Licensing You Need? Save Up to 20% by Chatting with our Experts!

Get Expert Licensing Help

A cloud security assessment checklist is a repeatable set of checks that shows where your cloud environment is weak, where it’s under attack, and which weak points an attacker can actually reach. Most checklists are just audit-type questions. Is MFA on in Microsoft Entra ID? Is logging enabled in Azure Monitor? Are patches up to date on Windows Server? They’re valid questions, but they don’t tell you where in Microsoft’s tooling the answer lives, or which ‘no’ to worry about first.

Key Takeaways

  • A cloud security assessment checklist for Azure and multi-cloud environments covers 12 checks across three lenses (posture, threat protection and exposure), each mapped to a Microsoft Defender for Cloud capability.
  • Free foundational CSPM provides Secure Score, recommendations, and asset inventory. Attack path analysis, agentless scanning, and data security posture management require the paid Defender CSPM plan.
  • Microsoft’s 2026 Digital Defense Report puts the average time until an exposed cloud workload is attacked at 5.3 hours, so internet exposure belongs on every cloud security audit checklist.
  • Defender CSPM’s permissions management (CIEM) identifies over-permissioned and dormant identities across Microsoft Entra ID, AWS IAM, and Google Cloud IAM.
  • Defender for Cloud’s data security posture management imports sensitive information types and sensitivity labels from Microsoft Purview.
  • TrustedTech’s Cloud Security Envisioning Workshop runs Defender for Cloud against a production Azure environment for two to three weeks and rates the findings low, medium, or severe.

This checklist is built the other way round. Each of the 12 checks names the Microsoft Defender for Cloud capability that provides the answer, grouped into the three lenses TrustedTech uses in the Cloud Security Envisioning Workshop: posture (what’s weak), threat protection (what’s being attacked), and exposure (which weaknesses an attacker can chain together to reach your data).

It’s for IT security admins, cloud architects, and the CISOs/CIOs who sign off. It’s Azure-focused, with notes on AWS and GCP. Use it as a cloud security audit checklist you run every quarter, or as pre-work before a formal assessment.

Why does a cloud security checklist need three lenses?

A cloud security checklist requires three lenses: posture, threat protection, and exposure, each answering a very different question. Most Azure security reviews stop at posture because it’s the easiest thing to quantify. But a posture checklist treats every weakness as equal, and that’s not how an attacker thinks.

Threat protection tells you whether someone is already probing what you’ve built. Exposure analysis shows which weaknesses are reachable from the internet and how they chain together into a path to something valuable. Skip either one and you’ll fix the most obvious finding rather than the riskiest one.

What does the cloud security checklist cover at a glance?

The checklist is 12 checks, each one mapping to a Defender for Cloud capability and the plan that provides it.

# Check Where it surfaces Plan it depends on
1 Baseline Secure Score and recommendations Defender for Cloud posture dashboard Foundational CSPM (free)
2 Defender plan coverage per subscription Environment settings All plans
3 Server vulnerabilities and exposed secrets Agentless machine scanning Defender CSPM or Defender for Servers Plan 2
4 Over-permissioned and inactive identities Permissions management (CIEM) Defender CSPM
5 Server threat detection and endpoint protection Defender for Servers, Defender for Endpoint Defender for Servers
6 Malware in cloud storage Defender for Storage malware scanning Defender for Storage
7 Databases, containers and App Service Workload-specific Defender plans Databases, Containers, App Service
8 AI workload threats Threat protection for AI services Defender for AI Services
9 Where sensitive data lives Data security posture management Defender CSPM or Defender for Storage
10 Attack paths to critical assets Attack path analysis Defender CSPM
11 Hunting and exposure metrics Cloud Security Explorer, Security Exposure Management Defender CSPM, Security Exposure Management
12 A remediation plan with owners and dates Results and next steps Process, not a product


Lens 1: What are the posture checks?

1. How do you baseline your Secure Score and read the recommendations behind it?

Start with Secure Score, but don’t stop there. Secure Score is Defender for Cloud’s posture score for the resources it can see, so treat the first score as a baseline rather than a grade. The Microsoft Learn module “Identify security risks by using Cloud Security Posture Management” favors risk-based prioritization over working down a flat list of recommendations.

Cloud Security Posture Management (CSPM) is the ongoing assessment of your cloud configuration against security best practices. Defender for Cloud’s foundational CSPM tier is free and includes Secure Score, security recommendations, asset inventory, and assessment against the Microsoft Cloud Security Benchmark.

Red flag: a high-severity recommendation that’s been open for months with no owner.

2. Which Defender plans are actually on, subscription by subscription?

Plan coverage is the check people skip, and it’s what makes the Secure Score credible. Defender for Cloud is a set of plans (CSPM, Servers, App Service, Databases, Storage, Containers, AI Services, Key Vault, Resource Manager, and APIs), each enabled or disabled per Azure subscription.

In the Cloud Security Envisioning Workshop, we record the current state and the decision for every plan on every in-scope subscription in a design decision sheet. It sounds pedestrian, but it’s how you catch a production Azure subscription that was never onboarded to Defender for Cloud.

Red flag: a healthy Secure Score because your most important subscriptions aren’t connected.

3. How do you find server vulnerabilities and plain-text secrets without installing agents?

Agentless machine scanning takes snapshots of the VM disks and analyzes the OS and file system out of band, with no agent on the VM and no performance impact on the machine. It reports software inventory, vulnerabilities (via Microsoft Defender Vulnerability Management), plaintext secrets, and whether an EDR is present and correctly configured.

It works for Azure VMs, AWS EC2 instances, and GCP compute instances once they’re connected, and is available in Defender CSPM and Defender for Servers Plan 2.

Red flag: credentials in plain text on a VM that can access production data in Azure SQL Database.

4. How do you review identities as if they were infrastructure?

Review identities with Cloud Infrastructure Entitlement Management (CIEM), which involves identifying excess, unused, or misconfigured permissions. In Defender for Cloud, CIEM covers human and non-human identities, inactive and guest accounts that still have access, IAM users without MFA enforced, and administrative rights that aren’t restricted to a specific group.

From Microsoft’s 2025 Digital Defense Report: over 97% of identity attacks are password attacks, and MFA can block over 99% of identity-based attacks. For the Microsoft 365 side, see TrustedTech’s guides to MFA for Microsoft 365 teams and why identity and access is the weakest point in cloud security.

CIEM covers Entra ID, AWS IAM, and GCP IAM as part of the Defender CSPM plan. Overprovisioned identities are the ‘stepping stones’ in the attack paths you’ll look at in check 10.

Red flag: a Microsoft Entra service principal with extensive permissions that hasn’t been used in months.

Lens 2: What are the threat protection checks?

5. How do you verify server threat detection and sort out your endpoint protection overlap?

Defender for Servers automatically deploys Microsoft Defender for Endpoint, which is a definite plus. It gets complicated if the server already runs a third-party antivirus or EDR: running Defender for Endpoint in parallel with a third-party AV/EDR can lead to performance and compatibility issues, as well as gaps in the threat signal.

Decide which product is in charge for each Windows Server and Linux server group. That’s why the Cloud Security Envisioning Workshop kickoff covers Microsoft Defender for Endpoint compatibility.

Red flag: two security agents fighting over the same server, or a server with neither working properly.

6. How do you scan cloud storage for malware, and what are the limits?

Defender for Storage scans Azure blobs for malware when malware scanning is enabled (on upload/modification, or on demand for existing blobs). It uses Microsoft Defender Antivirus, the same engine and signatures as Defender for Endpoint, and scans archives (e.g. ZIP, RAR). It matters most where users, customers, or partners upload files to Azure Blob Storage.

Know the limits. Scanning is static analysis with very limited emulation, so it will miss threats an endpoint scan would catch. Scan results are written as blob index tags, which aren’t tamper-resistant, so don’t rely on them as your only control.

Red flag: an upload pipeline feeding an Azure storage account with no malware scanning.

7. How do you check coverage for databases, containers, and App Service?

Databases, containers, and App Service each have their own Defender plan, and each is a separate yes/no on your checklist.

The point isn’t to turn everything on; it’s to make a considered decision. Each Defender plan has a cost, so the right answer depends on which workloads are internet-facing, which contain regulated data, and which could be retired. TrustedTech’s Microsoft Defender FAQ covers licensing and cost in more detail.

Red flag: an internet-facing web app or production database that nobody intended to leave unprotected.

8. Why should AI workloads be part of the review?

AI workloads belong in the review because your attack surface now includes the prompts and responses from services such as Azure OpenAI. The IBM 2025 Cost of a Data Breach Report (by the Ponemon Institute across 600 organizations) found that 13% of organizations had experienced a breach of an AI model or application, and of those, 97% reported having no AI access controls.

Threat protection for AI services in Defender for Cloud detects threats to generative AI services (jailbreaks, data leakage, data poisoning, credential theft, etc). It uses Azure AI Content Safety Prompt Shields and Microsoft threat intelligence, and alerts are fed into Defender XDR, where they can be correlated with other incidents. Protection covers text tokens only. For the governance side, see TrustedTech’s guide to agentic AI security risks.

Red flag: an AI app in production that no one on the security team has looked at.

Lens 3: What are the exposure checks?

9. Where does sensitive data live, and are you using your Purview definitions?

Defender for Cloud answers the first question with Data Security Posture Management (DSPM), which discovers sensitive data across your Azure storage and databases, including shadow data stores you’ve never inventoried. DSPM samples files rather than scanning every file, so Microsoft recommends using Microsoft Purview scanning if you need full cataloging.

This is where Defender and Purview come together. Defender for Cloud uses the built-in sensitive information types from Microsoft Purview, and you can import your own custom types and sensitivity labels to keep classification consistent. TrustedTech’s Microsoft Data Security Envisioning Workshop guide covers the Purview side in depth.

Red flag: a publicly accessible Azure Storage account that contains sensitive data.

10. How do you trace attack paths to your critical assets?

Trace them using attack path analysis in the Defender CSPM plan. Microsoft’s 2026 Digital Defense Report shows the average time before an exposed cloud workload was attacked was 5.3 hours, which makes the paths from the internet to your data the most time-critical part of the checklist.

Attack path analysis builds a graph of your environment and works out how an attacker could move from an external entry point (e.g. a VM exposed to the internet), through permissions, to a target (e.g. a database with sensitive data). It then recommends fixes that break the paths. It requires the Defender CSPM plan plus either agentless scanning or the vulnerability assessment capability in Defender for Servers, which is why checks 2 and 3 come first.

The practical benefit is finding choke points where a single fix breaks multiple attack paths.

Red flag: a path from the internet to sensitive data in 3 steps or less.

11. How do you hunt for what the dashboards don’t show, then measure it?

Hunt with Cloud Security Explorer, which sits on the same graph as attack paths, using the query builder to ask your own questions. For instance: which publicly accessible data resources contain sensitive data, and which identities can reach them? Run a few of these queries before you accept that the recommendations list is the whole story.

Microsoft Security Exposure Management adds a further layer. It covers attack paths, exposure metrics, and security initiatives, and maps exposures to real-world threats and business-critical assets, so you can measure whether your exposure is decreasing.

Red flag: no way to show your CISO whether exposure is going up or down quarter-on-quarter.

12. How do you turn the findings into a plan with owners and dates?

Give each finding a severity, a recommended mitigation, a named owner, and an expected date. A checklist that ends in a spreadsheet of findings hasn’t finished the job. In the Cloud Security Envisioning Workshop results, we rate findings as low, medium, or severe and conclude with a next-steps table that covers the action, the date, and notes.

Red flag: same Defender for Cloud findings on the next review, unchanged.

Which cloud security guidelines sit behind the checklist?

Five cloud security guidelines sit behind the 12 checks, and they apply to any Azure or multi-cloud environment:

  • Connect everything before you score anything. A partial view flatters your Secure Score.
  • Prioritize by reachability, not label. A medium-severity issue on an internet-facing path can outrank a high-severity one buried deep.
  • Treat identities as infrastructure. Permissions are often the quickest way to your data.
  • Classify your data once. Reuse your Microsoft Purview definitions so ‘sensitive’ means the same thing everywhere.
  • Decide on purpose. Every Defender plan on or off should be a deliberate decision, not an accident.

How do you run the checklist on your own tenant?

Run it as part of the Cloud Security Envisioning Workshop, which is designed to bridge the gap between a checklist and your actual environment. A checklist tells you what to look for, not what’s in your Azure tenant.

The engagement enables Microsoft Defender for Cloud in your production environment, so the findings come from your actual telemetry rather than a demo tenant. The process works as follows:

  1. Scoping: TrustedTech and your team agree on the Azure subscriptions, resources, and workload modules to include.
  2. Setup: The mandatory modules are cloud security posture management and the Microsoft Defender portal, followed by the workload modules you’ve selected.
  3. Data collection: Defender for Cloud collects telemetry for 2-3 weeks, with a check-in halfway through to confirm data is flowing and flag any critical findings.
  4. Exploration: TrustedTech and your team go through the posture, threat protection, and exposure findings together.
  5. Results and next steps: You get a prioritized set of findings, recommended mitigations, and an action plan. Decommissioning (removing the configuration afterward) is optional.

More information on the Cloud Security Envisioning Workshop and its modules is available. You can also see how TrustedTech approaches Microsoft Defender for Cloud, watch the Protecting Every Layer webinar, or read about the Modern SecOps Envisioning Workshop for Microsoft 365 environments.

Frequently Asked Questions

Q. What should a cloud security assessment checklist contain?

A. Security posture (misconfigurations, plan coverage, identities), threat protection for your workloads, exposure analysis (attack paths, sensitive data discovery), and a remediation plan with named owners. The checklist above covers all four in 12 checks.

Q. How often should we be running a cloud security audit checklist?

A. Quarterly, and again after any significant change such as a new subscription, a migration or a new AI workload. Your regulatory requirements may dictate a different frequency.

Q. How does Microsoft Purview fit into cloud security?

A. Microsoft Purview provides the sensitive information types and sensitivity labels that Defender for Cloud’s data security posture management uses, so classification stays consistent. For full data cataloging, Microsoft recommends running Purview scans alongside Defender for Cloud’s sampling-based discovery.

Q. Will a workshop trigger Defender for Cloud charges?

A. Defender for Cloud plans offer a 30-day free trial per subscription and per plan where a trial is still available. The workshop does not offset Defender costs, and resources become billable once the trial expires or where no trial applies. TrustedTech confirms trial availability at scoping and can estimate costs before anything is enabled.

Where are we going from here?

This cloud security assessment checklist works on its own. Four steps get you started:

  1. Connect up every Azure subscription and verify Defender plan coverage (checks 1 and 2).
  2. Close the red flags in posture and threat protection (checks 3 to 8).
  3. Trace exposure across sensitive data, attack paths, and hunting queries (checks 9 through 11).
  4. Assign owners and dates (check 12), and rerun the checklist next quarter.

If you’d rather see how your environment answers all 12 questions, talk to TrustedTech about a Cloud Security Envisioning Workshop, and we’ll help you scope the right modules for your environment.

Thomas Rosquin, Sr Writer

Thomas Rosquin, Sr Writer

Thomas Rosquin is a content strategist and technology writer at TrustedTech, a top 1% global Microsoft Cloud Solution Provider. With 20 years of experience in research, editorial, and content strategy, he focuses on Microsoft technologies, workplace AI, and IT governance, translating complex licensing and adoption decisions into clear guidance for technology leaders. His work draws on original research, industry analysis, and close collaboration with TrustedTech's Microsoft-certified solutions team.

LinkedIn | Case Studies