Skip to content
Back to overview Illustration: a falcon-shaped sensor watches over cloud blocks and servers, individual misconfigurations glow orange
CrowdStrike Azure Cloud Security

Connecting CrowdStrike Falcon Cloud Security to Azure: CSPM for subscriptions and Entra ID

By , Security Engineer 9 min read

Key takeaways

  • Falcon Cloud Security reads Azure configurations through the Azure APIs, reports misconfigurations as indicators of misconfiguration (IOMs) and can detect suspicious behaviour from activity logs and Entra logs.
  • Registration happens at management group or individual subscription level. CrowdStrike mainly receives read access in Azure and read permissions for Microsoft Graph. An additional custom role for App Service, however, also allows reading app settings and connection strings, according to the template. Review this before you approve it.
  • Setting it up requires Owner rights on the registered scopes and an Entra role that is allowed to grant Graph application permissions. Treat it as a planned change, not something done in passing.
  • A CSPM tool finds problems; it does not fix them. Before you start, decide who handles which findings within which deadline and how exceptions are documented.

Many organisations already run CrowdStrike Falcon as their EDR on clients and servers and ask whether the same platform can assess their Azure environment. It can, through the Falcon Cloud Security module. Little has been published about it outside the vendor documentation, most of which sits behind the customer login. This article explains what the connection does, which permissions it creates in your tenant and how to turn findings into a process you can follow and evidence. For management, the last point matters most: an auditor does not ask whether a tool reports findings, but what happened to them.

What does Falcon Cloud Security do in Azure?

Falcon Cloud Security is CrowdStrike’s cloud security platform. The vendor positions it as a CNAPP, a platform that combines several assessment areas. Three of them matter most for an Azure connection:

  • CSPM (cloud security posture management): continuous assessment of resource configurations such as storage accounts, network security groups or key vaults. CrowdStrike reports deviations as IOMs with a severity and remediation guidance.
  • CIEM (cloud infrastructure entitlement management): analysis of identities and permissions. According to CrowdStrike, the module monitors users, groups and application registrations in Microsoft Entra ID, shows privileged access and highlights excessive permissions or missing MFA.
  • CDR (cloud detection and response): detection of suspicious activity in the cloud control plane, for example unusual role assignments. CrowdStrike calls these indicators of attack (IOAs). They are based on the subscriptions’ activity logs and the Entra ID sign-in and audit logs.

On top of this come compliance assessments against frameworks such as CIS, NIST or PCI DSS as well as custom frameworks, and a graph-based analysis that links misconfigurations, vulnerabilities and excessive permissions into potential attack paths. Which of these capabilities you have depends on what you license. Clarify that with CrowdStrike or your partner before planning.

The practical benefit for organisations already using Falcon as EDR is that cloud findings and detections from the Falcon sensor on virtual machines end up in the same console. The security team does not need to learn a second tool for cloud findings.

How does registering an Azure tenant work?

Azure is hierarchical: below the Entra tenant sit management groups, and below them the subscriptions that contain the actual resources. Falcon Cloud Security can be registered at two levels. If you register a management group, Falcon discovers all subscriptions below it automatically, including ones added later. Alternatively, you register individual subscriptions. For environments built on a landing zone, we recommend registering the top-level management group so that no new subscription appears outside the assessment.

CrowdStrike publishes a Bicep template and a Terraform module on GitHub for the setup. According to the public documentation of these templates, the main building blocks are:

ComponentPurposePermission
App registration with service principalRead access for CrowdStrike to the Azure resources”Reader” role on the registered management groups or subscriptions, plus a custom role for App Service with the actions Microsoft.Web/sites/config/list/Action and Microsoft.Web/sites/publish/action (see below)
Microsoft Graph permissionsAssessment of Entra IDRead permissions; the Terraform module’s example lists User.Read.All, Policy.Read.All, RoleManagement.Read.All and AuditLog.Read.All, among others
Event Hubs in an infrastructure subscriptionReal-time visibility of activityDiagnostic settings forward the subscriptions’ activity logs and the Entra ID sign-in and audit logs
Optional: agentless scanning resourcesVulnerability and data scansDedicated scanner roles, managed identity and key vault

The App Service role deserves a close look. According to Microsoft, the action Microsoft.Web/sites/config/list/Action lists a web app’s security-sensitive settings, such as publishing credentials, app settings and connection strings. Microsoft describes Microsoft.Web/sites/publish/action as “Publish a Web App”. If app settings hold passwords or keys in plain text rather than as Key Vault references, the CrowdStrike service principal can read them. Ask CrowdStrike what the actions are needed for and decide deliberately whether to accept the role as it is. Either way, secrets belong in Azure Key Vault.

Whoever runs the registration, person or pipeline, needs several things. The Bicep template requires the Owner role on the management groups and subscriptions being onboarded, because it creates role assignments. On the Falcon side, an API client with the “Cloud security Azure registration” scope is needed. An administrator has to grant the Graph application permissions; according to Microsoft, only a Privileged Role Administrator or a more privileged role can do this for Microsoft Graph application permissions. Diagnostic settings for Entra ID require at least the Security Administrator role, and Microsoft requires an Entra ID P1 or P2 licence for exporting sign-in logs.

Our recommendation: run the registration with Terraform or Bicep from a pipeline rather than by hand in the portal. The Git history then documents which scope was connected, when and with which rights. The Falcon API client secret belongs in a key vault or the pipeline’s secret store, not in a variables file. After setup, compare the role assignments in the portal under Management groups > Access control (IAM) and the granted permissions in the Entra admin center under Entra ID > Enterprise apps > [application] > Permissions against the documentation.

Which findings does the connection deliver?

After the first assessment, the Falcon console shows three kinds of results, and each is handled differently.

Misconfigurations (IOMs) are states: a storage account allows public access, a network security group exposes management ports to the internet, logging is missing. They stay open until someone changes the configuration. Prioritisation and deadlines are what count here.

Behaviour-based detections (IOAs) are events: an identity grants itself unusual rights, a service is created in an unusual place. These belong in the incident process, not in the configuration backlog. Without real-time visibility through the Event Hubs, they are missing.

Identity findings concern Entra ID and Azure roles: permanently privileged accounts, app registrations with far-reaching Graph permissions, accounts without MFA. In assessments we often see the same patterns here, such as permanent Global Administrators or old app registrations with client secrets that nobody can attribute any more. These findings often have the widest impact because one compromised identity reaches many resources at once.

Who fixes the findings?

A CSPM tool creates work for several teams. Without agreed ownership, the list of open findings grows, and in an audit you cannot show that the organisation knew about the risks and dealt with them. A simple split has proven effective:

  • The security team (internal or a provider) reviews new findings, rates severity in context and assigns them. It does not decide every change, but it tracks deadlines.
  • Workload owners fix misconfigurations in their subscriptions. If the environment is managed as infrastructure as code, the fix goes into the code; otherwise the finding returns with the next deployment.
  • The platform team closes recurring classes of error centrally, for example with Azure Policy and the “deny” effect on the management group.
  • Management or the information security officer decides on exceptions that knowingly accept a risk and signs them off.

Set a deadline per severity and an escalation step for when it is missed. IOAs go into the incident process with on-call cover. If an external IT provider runs your Azure environment, remediation deadlines for findings belong in the contract; otherwise it remains unclear who has to act. How to check whether your provider fills that role is covered in a separate article.

How do findings become evidence?

For NIS2, DORA or ISO 27001 you do not need a perfect state, but a documented way of handling deviations. The Falcon connection produces several pieces of evidence:

  • The registration scope in the infrastructure-as-code repository shows which subscriptions are assessed.
  • Compliance assessments against CIS or NIST show the state on a given date. Export them regularly so that a history builds up.
  • Documented exceptions with justification, owner and expiry date show deliberate risk decisions.
  • Detections in Microsoft Sentinel, if you analyse centrally there. Microsoft offers a CrowdStrike connector for Sentinel that retrieves alerts, detections, hosts, cases and vulnerabilities through the CrowdStrike API. Check before the project whether the cloud findings that matter to you are included in these data types.

The difference from a screenshot is the history: an auditor can follow when a finding appeared, who worked on it and when it was closed.

Next step

Our Azure Cloud Security services cover connecting CSPM tools, the permission model behind them and the process for findings; this includes the Azure assessment against the Microsoft cloud security benchmark. If you first want an overview of Microsoft 365, Entra ID and Azure, our Security Assessment reviews your environment at a fixed price.

Book an introductory call

Frequently asked questions

Does Falcon Cloud Security need an agent in Azure?

Not for assessing configurations and identities. That runs through the Azure and Graph APIs. Runtime protection for virtual machines or containers, on the other hand, requires the Falcon sensor on those systems.

Which rights does CrowdStrike get in our tenant?

In the base configuration, mostly read rights: the Azure Reader role on the registered scopes and read permissions for Microsoft Graph. In addition, there is a custom role for App Service that can also read app settings and connection strings. Agentless scanning adds further roles. Review all roles before enabling them and document the decision.

Are new subscriptions picked up automatically?

Yes, if you register a management group. Falcon then discovers all subscriptions below it. Make sure new subscriptions in your landing zone are actually created under that management group.

Does Falcon Cloud Security replace Microsoft Defender for Cloud?

Not necessarily. The free Foundational CSPM tier of Defender for Cloud remains available in every subscription. Whether you also need a paid Microsoft CSPM plan depends on which tool your security team treats as the lead and where licences already exist.

Let's talk about your IT security

In the introductory call (30 minutes, no obligation) we clarify:

  • where your company stands today and what concerns you most
  • which next step makes sense for your organisation
  • what effort and price range to expect
Bastian Wenske

Bastian Wenske

Founder & Managing Director

Heidelberger Straße 15
12059 Berlin
Germany

Prefer to reach out directly? Send us an email at [email protected].