By Muhammad Ahsan | Solution Savy | Microsoft Defender Deep Dive, Part 5

Onboarding devices and configuring attack surface reduction rules establish an important defensive baseline. However, prevention does not remove the need to investigate suspicious activity. Microsoft Defender for Endpoint provides endpoint detection and response (EDR), live response, and automated investigation and remediation (AIR) to help security teams understand what happened and take appropriate action.

These capabilities serve different purposes. EDR supplies detections and investigation context. Live response provides a remote investigation shell for supported devices. AIR investigates supported alerts and can remediate identified threats according to the configured automation level. This article explains how to combine them within Microsoft Defender XDR without treating every alert as proof of compromise or every remediation action as closure.

1. Establish response readiness before an incident

Start by checking entitlement and device coverage. The capabilities discussed here are associated with Microsoft Defender for Endpoint Plan 2; do not assume that onboarding under Plan 1 provides the same investigation and response features. Also check the supported operating systems and prerequisites for each response action. A control available on one platform is not necessarily available, or equivalent, on another.

In the Microsoft Defender portal, review the device inventory and open representative device pages. Confirm that expected devices are reporting, investigate sensor health issues and verify that analysts can access the correct device groups. An onboarded device that is not communicating cannot be treated as immediately available for remote response.

Define who can investigate, isolate devices, approve remediation and use live response. Apply the appropriate permissions through the role-based access control model active in your tenant, including Microsoft Defender XDR unified RBAC where enabled.

Configure device groups deliberately. Alongside their access-control role, device groups determine the automation level applied to their member devices.

Enable live response in the Microsoft Defender for Endpoint advanced features settings when required. Check the additional platform-specific prerequisites before assuming that every device supports it.

Document operational ownership as well as technical access. The security team may authorise containment, while a service owner assesses business impact and an infrastructure team performs recovery. Practise these hand-offs on a test device. A role assignment is not a substitute for an agreed decision process.

2. Investigate the incident, not just the endpoint alert

Begin in the Microsoft Defender XDR incident queue. An incident can bring together related alerts and evidence across supported workloads. Examine the incident’s scope before moving into the endpoint details: which devices, users and other entities are involved, and what relationship connects the alerts? This reduces the risk of treating one endpoint detection as an isolated event.

Open the relevant alert and review its evidence, process relationships and affected entities. Then use the device page and timeline to examine activity around the detection. Look for the initiating process, command-line arguments, file locations, account context and subsequent activity. A suspicious process name alone is a weak basis for containment; its execution chain and surrounding behaviour provide stronger context.

Advanced hunting is useful when you need a repeatable view of process activity. The following query retrieves process creation events for a specific device over the previous 24 hours. Replace the placeholder with the actual device ID, and narrow the time range around the alert when possible.

DeviceProcessEvents
| where Timestamp > ago(24h)
| where DeviceId == "REPLACE_WITH_DEVICE_ID"
| project Timestamp,
          DeviceName,
          AccountDomain,
          AccountName,
          FileName,
          ProcessCommandLine,
          SHA1,
          InitiatingProcessFileName,
          InitiatingProcessCommandLine
| order by Timestamp asc

DeviceProcessEvents contains process creation and related events from Microsoft Defender for Endpoint. This query is a starting point, not a complete forensic record. Its results depend on available telemetry, the selected period and your access scope. No returned rows do not prove that no execution occurred.

Build a short chronology: initial execution, suspicious follow-on activity, detection and any response already taken.

Separate observed facts from hypotheses. Record what the telemetry shows and what still needs verification.

Expand the investigation to other devices or identities when the incident evidence justifies it, rather than assuming the original device is the entire scope.

3. Select containment actions proportionate to the evidence

Microsoft Defender for Endpoint exposes device response actions in the Microsoft Defender portal, with availability depending on platform, permissions and prerequisites. These include isolating a device, collecting an investigation package and running an antivirus scan. Use the action that addresses the immediate risk while preserving the information needed for investigation.

Device isolation restricts network communication while retaining connectivity to the Microsoft Defender for Endpoint service. This makes it valuable when there is credible evidence of active malicious activity. Nevertheless, assess supported isolation behaviour and connectivity requirements for the device’s platform. Network configurations, including VPN routing, can affect the service connectivity needed during isolation.

For credible active compromise, consider isolation promptly and record why the operational impact is justified.

Collect an investigation package when the supported collection can help your analysis. Treat the resulting package as sensitive evidence and store it according to organisational requirements.

Use an antivirus scan to support investigation and remediation, not as the sole test for whether a device is safe.

EDR in block mode is a separate consideration. With the required licensing and prerequisites, it allows Microsoft Defender for Endpoint to remediate malicious artefacts detected by EDR even when Microsoft Defender Antivirus is running in passive mode behind another antivirus product. Microsoft Defender Antivirus must remain installed and its cloud-delivered protection must be enabled.

EDR in block mode is not a replacement for active antivirus protection. It also does not mean every EDR detection results in a block. Treat it as an additional post-breach protection capability, and verify its prerequisites rather than assuming that EDR onboarding alone enables this behaviour.

4. Use live response as a controlled investigation capability

Live response provides a remote shell connection to a supported device through the Microsoft Defender portal. It is designed for investigation and response tasks, including examining processes and files, retrieving evidence and running approved scripts. It is not an unrestricted interactive PowerShell terminal: it uses its own command environment, with capabilities governed by permissions and platform support.

Start with a specific question. For example: does the suspicious file still exist, is the associated process still running, or is additional local evidence required? Open the device page, start a live response session and use the supported commands needed to answer that question. Use the built-in help rather than assuming that ordinary operating-system commands are available.

Perform read-only inspection first where circumstances permit. Record relevant paths, process details and timestamps before making changes.

Retrieve only the evidence necessary for the investigation. Live response file retrieval can expose sensitive organisational or personal information.

Use scripts from the live response library only after review and testing. The ability to upload and execute scripts should be restricted to appropriately authorised responders.

Avoid enabling unsigned script execution merely for convenience. Where it is required, document the exception and apply additional approval and review.

Live response maintains session and command information that supports review. Reference that information in the incident record, together with your findings and changes. Do not rely solely on an analyst’s recollection of what happened during the session.

For example, if suspicious PowerShell activity appears in the timeline, first inspect the execution chain and available command-line telemetry. Use live response only when additional device-level evidence is necessary. If a reviewed script is needed, define its expected output and potential side effects before execution. This keeps remote access tied to an investigation objective rather than exploratory changes.

5. Govern automated investigation and remediation

AIR uses investigation logic to examine supported alerts and related evidence. It can investigate entities such as files and processes, determine whether remediation is necessary and take or propose supported remediation actions. It reduces repetitive investigation work, but it does not replace an analyst’s assessment of the wider incident.

The configured device group automation level determines how remediation proceeds. Full automation allows identified threats to be remediated automatically. Semi-automated levels require approval for specified remediation circumstances. Microsoft recommends full automation; organisations should still validate device grouping, permissions and operational responsibilities before relying on it.

Use a representative test group to understand the workflow. Review investigation results, the evidence supporting proposed actions and the business effect of remediation. If approval-based automation is selected, assign explicit ownership for pending actions. An approval queue without an accountable reviewer leaves the response workflow incomplete.

Review the automated investigation’s scope and findings alongside the incident evidence. Do not assume that an investigation assessed every possible route of compromise.

In the Action centre, review pending actions and approve or reject them based on the evidence and your response authority.

Check action history to distinguish proposed remediation from completed remediation. Where an action supports reversal, use that capability only after reviewing the consequences.

A completed automated investigation is not the same as a closed incident. Confirm that relevant remediation actions completed, investigate any remaining suspicious activity and address the entry point. If the evidence also indicates identity compromise or activity on other devices, endpoint remediation alone does not resolve that broader scope.

6. Short operational checklist

Readiness: verify Plan 2 entitlement, platform support, sensor health, device groups and response permissions.

Investigation: review incident relationships, alert evidence, the device timeline and targeted hunting results.

Containment: choose supported actions, document their justification and assess business impact.

Live response: define the question, favour read-only collection and use reviewed scripts with appropriate permissions.

Automation: confirm the applicable automation level, assign pending-action ownership and verify completed actions.

Closure: establish incident scope, record evidence, address the entry point and agree recovery and monitoring with the service owner.

7. Conclusion: make response verifiable

Effective endpoint response is a sequence of evidence-based decisions. Use EDR to understand suspicious activity, device actions to contain it, live response to answer specific investigative questions and AIR to handle supported investigation and remediation tasks under an agreed automation model.

The operational standard should be straightforward: responders can explain what they observed, why they acted and how they verified the result. That is more valuable than enabling every response capability without clear ownership. In Part 6, the series moves to Microsoft Defender for Identity and the investigation of attacks involving on-premises Active Directory.

Microsoft Learn: Take response actions on a device

Microsoft Learn: DeviceProcessEvents advanced hunting table

Microsoft Learn: Endpoint detection and response in block mode

Microsoft Learn: Investigate entities on devices using live response

Microsoft Learn: Automated investigation and remediation

Microsoft Learn: Automation levels in automated investigation and remediation


Originally published on Solution Savy. Need help with this? Book a consultation.