That Green Status Needs a Second Look
When an endpoint detection and response (EDR) platform reports successful isolation, the first task is confirming what that status actually means.
A response tool may acknowledge an isolation request before the operation has taken effect. Even a confirmed isolation can leave some communications permitted under the configured policy.
Start with the exact asset and action. Confirm that the intended system was isolated, identify any permitted connections and establish when the control became effective.
If malicious activity continued after the isolation request, the investigation needs to determine whether the action was delayed, incomplete or simply unable to affect the attacker's remaining access.
The team should also understand the business impact. Isolating an application server might interrupt a critical integration or prevent administrators from retrieving information needed during the investigation.
Those decisions need coordination between security and service owners.
It's also worth understanding how automated response decisions are triggered.
A predefined playbook and an AI-driven response may achieve the same containment outcome, but they introduce different questions about authorisation, evidence and oversight.
In either case, the case record should distinguish between the action requested, the result confirmed and the threat activity that remains possible.
The Attacker May Have Left the Host Behind
The asset that triggered an alert is often the easiest part of an incident to identify.
Understanding what the attacker did before that alert is considerably harder.
Suppose a cloud-hosted build runner is compromised. The security platform isolates the instance, preventing further activity through its ordinary network connections.
During the investigation, analysts discover that the attacker may have accessed a deployment credential.
That credential could provide access to other services, depending on its permissions and whether it remains valid.
The isolated instance is now only one part of the investigation.
The team needs to establish whether the credential was exposed, whether it was used elsewhere and which systems could have been reached through it.
Start with the evidence already available. Follow related identities, resources and events, expanding the investigation where the findings justify it.
For a cloud or identity-based incident, that may involve reviewing authentication records, service principals, application permissions and access to connected services.
Revoking Credentials May Leave Sessions to Investigate
Identity remediation needs its own verification. Disabling an account or revoking a token does not necessarily terminate every existing application session immediately.
The result depends on the identity provider, token type and application configuration.
For example, Microsoft's guidance on emergency access revocation explains that existing access tokens and application-managed sessions can remain relevant after certain identity actions.
Responders should confirm which access mechanisms were affected and what remains active.
The investigation may require additional session revocation, credential rotation or application-specific controls, coordinated with the teams that own those services.
Follow the investigation through the decisions and handoffs that determine how the incident ends.
Find Out How Far the Incident Travelled
Once the immediate threat is restricted, the investigation needs to establish its scope.
This means reconstructing the attacker's activity across the available evidence, identifying potentially affected assets and determining where the compromise may have spread.
Start with the earliest observed suspicious activity, then examine the sequence leading to containment.
Use the evidence to guide the search. A connection to another service may justify reviewing its authentication and application records. An unexpected credential change may lead investigators towards identity administration logs.
As our guide to the incident response lifecycle explains, containment is one part of a response that must also account for investigation, eradication and recovery.
A successful first action buys the team an opportunity to complete that work.
Don't Rebuild Away the Evidence
Containment and recovery actions can change or destroy information investigators still need.
Terminating a compromised cloud instance may remove volatile evidence. Rebuilding a system can eliminate local artefacts that would help explain how the attacker gained access or what they did afterwards.
Before destructive remediation, identify which evidence is needed and whether it can be collected safely.
That may include memory, relevant system and application records, cloud audit history and copies of malicious artefacts. Prioritise volatile information and sources with short retention periods where they matter to the case.
There is a trade-off to manage. Preserving evidence should not require allowing an active attack to continue unchecked. The incident commander needs the authority to balance the immediate risk against the investigative value of preserving information.
Give Each Team a Job That Moves the Incident Forward
Post-containment investigations can stall when several teams assume somebody else is handling the remaining work.
The SOC may be investigating the original alert while the identity team is waiting for a request to revoke access. Infrastructure engineers may be preparing to rebuild the system before forensics has identified the evidence required.
A clear division of responsibility helps prevent those gaps.
The roles will differ by organisation, but each important question needs someone accountable for answering it.
At shift changes, make unresolved work explicit. A task assigned to an analyst who has finished their shift can quietly disappear from active attention unless ownership is transferred.
The handover should explain what has been established, what remains uncertain and which decision is needed next.
Leadership Needs More Than an Isolation Timestamp
Executives and business owners may hear that a threat has been contained and reasonably assume the incident is nearing its end.
The security team needs to communicate the difference between restricting the immediate activity and establishing the wider consequences.
A useful incident update might say:
The affected cloud instance has been isolated, and the control is confirmed. Investigation of its associated deployment credentials is continuing because those credentials may have provided access to other services.
The identity team is reviewing relevant activity and permissions. Recovery of the affected workload remains pending until that access is assessed and required evidence is preserved.
This gives decision-makers a meaningful view of progress. It also helps them understand why service restoration or additional remediation may still be necessary.
Follow the investigation through the decisions and handoffs that determine how the incident ends.
Make the Next Automated Response Smarter
After the incident, review how well the original containment performed.
Did the rule trigger on the intended behaviour? Did it select the correct target? Was there a delay between the request and its confirmed result?
Also examine what happened around the automation. Perhaps the initial response worked, but nobody investigated an exposed service credential until the next shift. Perhaps the playbook correctly isolated a workload while creating an unexpected dependency for another business service.
Those findings can improve the response process.
A playbook may need an additional verification step, a clearer escalation trigger or an explicit handoff to identity and cloud teams.
Detection engineering may need to improve the evidence available before containment or add visibility into related activity.
Assign improvements to owners and retest the affected workflow before expanding automation.
Teams should be able to demonstrate that the revised response performs better, including what happens when the first action succeeds but the investigation remains incomplete.
The Most Important Work Starts After the First Win
Fast containment is valuable. It can interrupt harmful activity and give responders time to understand what they're dealing with.
That time is useful only when the team knows what to do next.
The strongest incident response teams can verify the initial action, investigate the wider compromise, coordinate remediation and make recovery decisions supported by evidence.
Those capabilities are difficult to judge from a dashboard showing successful automated actions. They're better understood when the team has to work through the full response together.
That's the thinking behind Live Breach Exercises, where security teams investigate realistic incidents and practise the decisions involved in containment, eradication and recovery within a controlled environment.
Follow the investigation through the decisions and handoffs that determine how the incident ends.
Ayomide Joseph A.