Blog Post

Linux and Open Source Blog
7 MIN READ

Now in public preview: DISA STIG benchmarks and per-rule auto-remediation for Linux

pallakatos's avatar
pallakatos
Icon for Microsoft rankMicrosoft
Nov 17, 2026

Audit your Linux machines against CIS and DISA STIG benchmarks, then automatically remediate selected settings. You control what changes and where remediation runs.

In our previous announcement, we shared that built-in CIS Benchmark auditing for Linux was now Generally Available. We also said that STIG benchmarks and granular auto-remediation were coming next.

Today we're excited to bring both into Public Preview: per-rule auto-remediation for CIS Benchmarks, and DISA STIG benchmarks with auditing and auto-remediation for supported rules.

When we introduced built-in CIS Benchmarks, we wanted to make Linux compliance easier to manage without taking away the flexibility customers need. This release continues that work through Azure Policy with Machine Configuration and the same customizable security baseline experience.

It works across supported Linux VMs in Azure and Azure Arc-enabled servers on-premises or in other clouds. CIS auditing remains GA; the new remediation capabilities and STIG support are in Public Preview.

CapabilityStatus
CIS Benchmarks auditing for LinuxGenerally available
CIS Benchmarks auto-remediation for supported rulesPublic preview
DISA STIG auditing for LinuxPublic preview
DISA STIG auto-remediation for supported rulesPublic preview

Jump to: Remediation controls | Get started | Arc onboarding

Relevant Microsoft Learn resources:

Helping you maintain your Linux security posture

With auditing in place, you can see which machines meet your baseline and which settings need attention. The next question is how to fix those findings and keep the settings in place as your environment changes.

That's what we wanted to address with auto-remediation. For supported rules you enable, Machine Configuration can apply the desired settings and correct drift through ongoing evaluation. Your teams don't have to keep returning to the same finding or maintain a separate hardening script for every control we already support.

At the same time, we know that changing configuration across a fleet needs care. We've built the workflow so you can start with a few rules on test resources, see the effect on your workloads, and expand when you're ready.

STIG benchmarks are now built in

This is an important addition for our US public sector, Department of Defense, government and contractor customers working against DISA requirements. You can now bring DISA Security Technical Implementation Guide (STIG) assessments into the same built-in experience as CIS, including remediation for supported rules.

The initial STIG coverage includes:

  • Red Hat Enterprise Linux 8 and 9
  • Oracle Linux 8 and 9
  • Ubuntu 22.04 LTS and 24.04 LTS

Select the benchmark that applies to your machines, customize the supported settings, and start assessing. You don't need to adopt a different workflow when moving between CIS and STIG.

Figure 1. Choose the STIG baselines and versions that match your Linux estate.

Auto-remediation is a per-rule decision

This is the part of the design we want to highlight. Remediation is controlled per rule, not just at the policy level. Changing the policy effect to DeployIfNotExists does not mean the entire benchmark will be remediated.

Rules start in Audit. Where remediation is available, you can change an individual rule to Remediate and leave the others in Audit. The policy configuration and the Remediation Location parameter then determine whether and where those changes can run.

Figure 2. Every rule starts in Audit. You choose Remediate for each supported rule.

SettingWhat it controls
Rule mode (Audit or Remediate)Which supported rules can change settings on a machine
Policy effect: DeployIfNotExistsTurns on the Azure Policy deployment that remediation needs
Remediation LocationWhich resource locations remediation applies to
Remediation taskDeploys the configuration to machines that were already noncompliant

The Azure Policy remediation task deploys the configuration to existing noncompliant resources. Machine Configuration then applies the settings for the rules you've selected.

Let's say you have test resources in one region. You can enable remediation there first, while continuing to audit the rest of the assignment. Once you've reviewed the results and workload behavior, add more locations or enable more rules. We wanted that progression to be part of the built-in workflow, rather than something you have to manage separately.

Getting started with remediation

1. Start with an audit assignment

Create a CIS or STIG baseline assignment in audit mode, choose your distributions and benchmark versions, and set any custom values you need. Let the first assessment finish, then review the noncompliant findings.

2. Edit the assignment in the portal

Go to Policy > Assignment Presets > Machine Configuration, open Assignments, select your assignment, and choose Edit settings. The existing settings editor lets you make these changes in place; you don't need to recreate the assignment.

Figure 3. Open an existing baseline assignment with Edit settings.

3. Choose the rules you want to remediate

Use the Remediatable filter to see the rules we currently support. Change the ones you want to automate from Audit to Remediate.

Figure 4. Audit and Remediate rules in the same baseline.

The summary shows what you're about to enable. In this example, 7 rules are configured for remediation and 149 remain in Audit.

Figure 5. Check how many rules are set to Remediate before you save.

4. Set the policy effect and remediation locations

Continue into the policy assignment flow:

  1. Set Effect to DeployIfNotExists.
  2. Choose where to begin using Remediation Location.
  3. Create a remediation task for existing noncompliant resources.
  4. Save the assignment.

Figure 6. Remediation Location appears with the other assignment parameters. This screenshot shows the audit effect. Select DeployIfNotExists to enable remediation.

5. Review the results and expand

Allow about 15 to 20 minutes for remediation to start. Timing depends on policy processing, agent evaluation and machine connectivity; this is not a completion guarantee.

The next assessment shows the updated compliance state. Review the changes and their effect on your workloads before expanding to more rules or locations.

Figure 7. Compliance state, configured mode and reasons for each rule.

All assessment reports are available in Azure Resource Graph. You can use the built-in reporting experience or query the results to build your own dashboards and track progress across your fleet.

Custom parameters work for remediation too

The flexibility we introduced for auditing carries over to remediation. Wherever a rule supports both custom parameters and remediation, the same configured value is used for both.

For example, if your company's approved policy sets password expiration to 200 days, you can audit against that value and remediate to it. You don't need to maintain a separate script with another copy of the same setting.

As with auditing, a value that differs from the published benchmark is a customization of your baseline. The result tells you whether the machine meets your configured requirement, not the unmodified benchmark.

Target assignments with Inclusion Tags

We've also added Inclusion Tags for resource selection. If your machines are already tagged, you can use those tags to target a baseline assignment rather than relying only on resource group boundaries.

One caveat: Inclusion Tags control the assignment, not a separate remediation scope. Remediation within that assignment still depends on the rules you enable, the policy effect and Remediation Location.

We're planning an additional parameter to let you control remediation by tag as well as location. If that would help you separate test and production rollout groups, please let us know. It isn't part of this preview yet.

Assign a baseline as you onboard with Azure Arc

We're also making it easier to get those assessments in place when machines join your managed environment. When creating an Azure Arc multicloud connector, the Solutions step now includes Security Baselines (preview), powered by Machine Configuration.

From there, you can assign a CIS or STIG baseline to supported resources being onboarded through the connector. The AWS example below brings Arc EC2 onboarding and baseline selection into the same setup flow, so you don't have to come back and configure the assessment separately.

Figure 8. Select Security Baselines in the Solutions step to assign a CIS or STIG baseline as supported EC2 instances are onboarded to Azure Arc.

The same remediation controls still apply. Onboarding with a baseline doesn't automatically enable fixes across all its rules.

Under the hood: kompli for both CIS and STIG

If you've followed the earlier posts, you'll recognize kompli, our open-source Linux compliance and remediation engine. We designed it to support more than one benchmark family, and STIG is now using that same foundation.

The augmentation pipeline maps benchmark controls to kompli's reusable atomic procedures. These implement the underlying checks and, where supported, the corrective actions. When a CIS control and a STIG control require the same system operation, we reuse the same implementation, while retaining the control mappings and expected values for each benchmark.

This is why we don't need a separate collection of remediation scripts for each standard. As we extend the procedures, we can use that work across the controls that depend on them.

At runtime, Azure Policy orchestrates the assignments, Machine Configuration delivers and evaluates the configuration on each machine, and kompli runs the checks and supported fixes.

The same attention to assessment parity

In our CIS work, we've put a lot of care into comparing kompli's assessment results with CIS-CAT Pro and minimizing discrepancies. We're taking the same approach with STIG.

We measure kompli against DISA's SCAP Compliance Checker (SCC) to identify differences and understand why they occur. That comparison matters: customers should be able to understand how our assessments relate to the reference tool, not just receive another compliance score.

Our goal is to minimize discrepancies and understand the remaining differences. This is assessment comparison, not a claim of DISA certification.

Public Preview: tell us what you need next

We're starting with a smaller set of remediations intentionally. We want customers to try the workflow, see how it fits their operational processes, and help us prioritize the controls we add next. Not every audited rule supports remediation today.

With the initial set, we estimate that remediation can help improve compliance scores by around 5 to 10%, depending on the distribution and starting configuration. The actual gain depends on which supported rules are noncompliant and which you enable. We will gradually expand coverage so more findings can be resolved through the same built-in experience.

Our customer-driven approach continues here. If there's a rule you'd like us to remediate, a custom parameter you need, or a rollout scenario we haven't covered, tell us in the comments or open an issue in the kompli repository. We welcome requests for additional remediations and custom parameters for both CIS and STIG, as well as feedback on assessments, distribution coverage, and the overall experience. We'd particularly like your feedback on tag-based remediation scope.

Try it on your test resources, review the results, and let us know what would make this useful across more of your environment. You can also explore kompli on GitHub.

Updated Nov 16, 2026
Version 1.0
No CommentsBe the first to comment