Microsoft 365 security baselines: SCuBA, CIS, Maester
In short
A Microsoft 365 security baseline is a published list of recommended tenant settings. CISA's SCuBA baselines and the CIS Microsoft 365 Foundations Benchmark are standards. ScubaGear, EIDSCA, ORCA and Maester are tools and check sets that test a tenant against them. Pick one standard as your primary baseline, then use the tools to measure it and catch settings that drift.
Search for how to harden a Microsoft 365 tenant and the same names keep coming up: SCuBA, ScubaGear, CIS, EIDSCA, ORCA and Maester. They aren't interchangeable. Two are standards, and the rest are tools or check sets that test a tenant against them. Each has its own maintainer and scope.
What is a Microsoft 365 security baseline?
A Microsoft 365 security baseline is a published list of recommended configuration settings for a tenant. You compare your settings against it to find and close gaps. It describes a target state and doesn't change anything by itself.
Three terms get mixed up:
- Baseline: a set of recommended settings, usually with a rationale for each. CISA's SCuBA baselines are an example.
- Benchmark: a baseline produced through a formal consensus process, with numbered versions that auditors can reference. The CIS Microsoft 365 Foundations Benchmark is the best-known example.
- Assessment tool: software that reads your tenant's settings and reports which recommendations pass or fail, such as ScubaGear, ORCA and Maester. EIDSCA is a check set that runs inside Maester.
Microsoft also uses the phrase for Intune security baselines, which configure Windows devices, Microsoft Edge and Microsoft 365 Apps. This post is about tenant configuration: identity, email, sharing and collaboration.
What is CISA SCuBA, and what does ScubaGear check?
SCuBA (Secure Cloud Business Applications) is a CISA project that publishes secure configuration baselines for Microsoft 365 and Google Workspace. ScubaGear is CISA's free tool that checks a Microsoft 365 tenant against those baselines. SCuBA is written for US federal civilian agencies, but any organization can use the baselines at no cost.
The Microsoft 365 baselines are split by product: Microsoft Entra ID, Exchange Online, Microsoft Defender and related security settings, SharePoint and OneDrive, Microsoft Teams, Power Platform and Power BI. Each policy has a stable ID, such as MS.AAD.1.1v1 ("Legacy authentication SHALL be blocked"). The wording follows RFC 2119, the internet standard for requirement keywords. SHALL is a requirement, and SHOULD is a strong recommendation you may depart from with good reason. For federal civilian agencies, Binding Operational Directive 25-01 makes a set of these configurations mandatory. For everyone else, they are guidance.
ScubaGear is an open-source PowerShell module that collects your tenant's settings, then evaluates them against rules written from the baseline documents. The rules run in Open Policy Agent, an open-source policy engine. Results come back as an HTML report plus JSON and CSV files.
Install-Module -Name ScubaGear
# Run the one-time dependency command from the README, then:
Invoke-SCuBA -ProductNames *
ScubaGear runs in Windows PowerShell. The baseline documents live in the same repository, so you can read each policy's rationale next to your results.
How is the CIS Microsoft 365 benchmark different from SCuBA?
The CIS Microsoft 365 Foundations Benchmark is a consensus standard from the Center for Internet Security, a nonprofit. SCuBA comes from a US government agency. CIS sorts its recommendations into levels and publishes numbered versions. Version 7.0.0 arrived in May 2026, and CIS's June 2026 Benchmarks update lists its new, relocated and removed recommendations.
What else sets it apart:
- Profiles. Each recommendation states its profile. Level 1 is a practical baseline with little impact on how people work. Level 2 adds defense in depth that can affect functionality. Recommendations also note the license tier they assume, such as E3 or E5.
- Audit and remediation steps. Each recommendation explains how to check the setting and how to fix it, often with admin center steps and PowerShell.
- Version-specific numbering. Recommendation numbers shift between versions, so always record which version you measured against.
- A document first. CIS SecureSuite members can assess a tenant against the current version with CIS's own assessment tool. Otherwise, use a tool that maps checks to CIS recommendations, such as Maester's CIS suite, and confirm which version it targets.
What are EIDSCA and ORCA?
EIDSCA and ORCA are specialist checks that go deeper than a general baseline: EIDSCA for Microsoft Entra ID settings and ORCA for email protection in Exchange Online.
EIDSCA
EIDSCA (Entra ID Security Config Analyzer) is a set of checks for Microsoft Entra ID tenant settings that mitigate common identity attack scenarios. It comes from the Microsoft Entra ID Attack and Defense Playbook, a community project, and you run it as a test suite in Maester. The checks cover authentication methods, default user permissions, guest access, app consent, password protection, smart lockout and Microsoft 365 group settings.
These settings decide who can grant apps and agents access to company data, and who from outside can join. Sprawlsight's free snapshot checks two of the same areas as ENTRA-030 (Users can't consent to applications on their own) and ENTRA-032 (Guest invitations are restricted).
ORCA and the configuration analyzer
ORCA (Office 365 Recommended Configuration Analyzer) is a PowerShell report for email protection. It checks Exchange Online Protection and Defender for Office 365 settings, including anti-spam, anti-phishing and anti-spoofing policies, Safe Links and Safe Attachments. Microsoft product managers wrote it, but it is a community tool, not an official Microsoft utility. You install the ORCA module from the PowerShell Gallery, connect to Exchange Online and run Get-ORCAReport, which opens an HTML report in your browser. It needs only read access, which the Global Reader role provides. Maester also includes ORCA's checks as a test suite.
Microsoft's official, built-in counterpart is the configuration analyzer in the Microsoft Defender portal, under Email & collaboration > Policies & rules > Threat policies > Configuration analyzer. It compares your threat policies to the Standard and Strict protection settings in the preset security policies. It lets you apply recommendations and keeps up to 90 days of drift history. The target values are documented in Microsoft's recommended settings for email and collaboration.
What is Maester, and how does it relate to the others?
Maester is an open-source test automation framework for Microsoft 365 security configuration. It is built on Pester, the PowerShell testing framework, and Microsoft Graph. Its community maintainers publish it on GitHub under the MIT license.
Maester ties the others together. It bundles its own tests with suites for EIDSCA, CISA SCuBA, CIS and ORCA. One run produces a single interactive HTML report with links to where each setting is fixed. Because it uses Pester, you can add your own tests.
Install-Module Pester -MinimumVersion 5.7.1 -MaximumVersion 5.7.1 -SkipPublisherCheck -Force -Scope CurrentUser
Install-Module Maester -Scope CurrentUser
md maester-tests; cd maester-tests
Install-MaesterTests
Connect-Maester
Invoke-Maester
Maester is built for automation. You can schedule it in GitHub Actions, Azure DevOps or Azure Automation and keep each run's results. Compare-MtTestResult shows which tests changed between runs, and Maester can send a summary by email, Teams or Slack after each run.
One caveat: Maester's suites cover only part of their source baselines. Its CISA tests cover the Entra ID, Exchange Online and SharePoint baselines. Each CIS test cites the benchmark version it follows, which can trail the latest release, so check it before you compare recommendation numbers. ScubaGear remains CISA's reference tool for SCuBA.
How do SCuBA, CIS, EIDSCA, ORCA and Maester compare?
SCuBA and CIS are the standards you measure against. ScubaGear, EIDSCA, ORCA and Maester are how you measure.
| Name | Type | Maintained by | Covers | How you run it | Best for |
|---|---|---|---|---|---|
| CISA SCuBA baselines | Standard | CISA | Entra ID, Exchange Online, Defender, SharePoint and OneDrive, Teams, Power Platform, Power BI | Read the baseline documents | A free baseline with stable policy IDs |
| ScubaGear | Tool | CISA, open source | Every SCuBA Microsoft 365 baseline | PowerShell module on Windows | A full report against SCuBA |
| CIS Microsoft 365 Foundations Benchmark | Standard | Center for Internet Security | Microsoft 365 admin settings, by level and license | Read the benchmark, test with a mapped tool | Audits that reference CIS |
| EIDSCA | Check set | Community (Entra ID Attack and Defense Playbook) | Microsoft Entra ID settings | Maester test suite | Identity configuration depth |
| ORCA | Tool | Community, written by Microsoft product managers | Exchange Online Protection, Defender for Office 365 | PowerShell module, or Maester | Email protection tuning |
| Maester | Test framework | Community, open source | Its own tests plus EIDSCA, CISA, CIS and ORCA suites | PowerShell and Pester, locally or in a pipeline | Continuous, automated checks |
Where do Secure Score and the Zero Trust Assessment fit?
Both are Microsoft's own views of posture that complement a baseline rather than replace it.
Microsoft Secure Score measures security posture in the Microsoft Defender portal. You earn points for completing recommended actions across Microsoft 365 services. Treat it as a prioritized to-do list with a number, not a baseline to cite control by control.
Microsoft's Zero Trust Assessment is an open-source PowerShell module that tests hundreds of configuration settings. Its checks follow the Zero Trust pillars and the Secure Future Initiative, Microsoft's company-wide security program, and draw on NIST, CISA and CIS guidance.
Which baseline should you use?
Pick one standard as your primary baseline, then add the specialist checks.
- Choose SCuBA if you want a free baseline with stable policy IDs and a free reference tool, or if you work with US federal agencies.
- Choose CIS if your auditors, customers or compliance framework already ask for it, or if you want Level 1 and Level 2 profiles to phase the work.
- Add EIDSCA and ORCA either way, for deeper identity and email checks.
- Measure continuously with Maester on a schedule, not once a year.
The two standards agree on the fundamentals: block legacy authentication, require MFA, and restrict external sharing and user consent to apps. Where they differ, decide once and write the decision down.
How do you use a baseline well?
Pick one baseline and version, measure read-only, record exceptions, fix what exposes data first and recheck on a schedule.
- Pick a primary baseline and version. Record it, for example CIS v7.0.0, so results stay comparable over time.
- Start read-only. Use a reader role wherever the tool supports one, and fix nothing until you see the full picture.
- Record exceptions. For each recommendation you won't meet, note the reason, any compensating control, an owner and a review date. SCuBA's SHOULD wording expects this.
- Fix by exposure. Start with settings that control who can reach data: sharing links, guest access, app consent and Conditional Access. Microsoft 365 Copilot and AI agents work within those same permissions. Our posts on Copilot policy gaps and the Microsoft 365 permissions audit go deeper.
- Track drift. Settings change after an assessment: a new admin, a support ticket, a product update. Rerun on a schedule and compare results.
If you'd rather not run and reconcile these tools yourself, Sprawlsight's full assessment runs the CISA SCuBA and Maester benchmark suites for you alongside its own checks. It shows what changed between weekly scans. To close policy gaps yourself, the Policy Sprawl field guide covers DLP, sensitivity labels and Conditional Access measured against SCuBA. If Copilot is why you're looking, start with whether your tenant is ready for Copilot.
Frequently asked questions
Is ScubaGear only for US government agencies?
No. CISA built SCuBA for federal civilian agencies, but its project page says any organization can use the baselines, and ScubaGear is free and open source on GitHub. Only federal civilian agencies are bound by CISA's Binding Operational Directive 25-01. For everyone else, the baselines are guidance you can adopt in full or in part.
Do I need both the CISA SCuBA baselines and the CIS benchmark?
Usually not. Pick one as your primary baseline, record which version you measure against, and document your exceptions to it. If a customer or auditor later asks for the other, map your existing results to it rather than starting over.
Does Maester replace ScubaGear?
Not entirely. Maester is built for scheduled, automated runs, but its CISA checks cover only the Entra ID, Exchange Online and SharePoint baselines. ScubaGear is CISA's reference tool and covers every SCuBA baseline for Microsoft 365, including Defender, Teams, Power Platform and Power BI. You can use Maester for continuous monitoring and ScubaGear when you need a full SCuBA report.
Is Microsoft Secure Score a security baseline?
Not in the same sense. Secure Score measures posture in the Microsoft Defender portal by awarding points for recommended actions, and you can mark actions as covered by another control or accept the risk. It is useful for prioritizing work, but it isn't a versioned list of settings you can cite control by control the way you can cite SCuBA or CIS.
How often should I check my tenant against a baseline?
Run a full assessment at least quarterly and after major changes, and automate the checks that matter most so they run weekly. Settings drift as admins, support tickets and product updates change them. Scheduled Maester runs and the drift history in the Defender configuration analyzer help you catch those changes.