Threat reportVulnerabilityTL-2026-2161
CISA Red Team Fully Compromises Two Critical Infrastructure Orgs via ADCS ESC1 and AzureHound Cloud Enumeration (AA26-237A)
CISA advisory AA26-237A, "A Tale of Two SOCs," documents two simultaneous authorized red team assessments — one against a Government Services and Facilities Sector organization, one against a Water and Wastewater Systems Sector organization — that used identical tradecraft (default credentials, phishing, ADCS ESC1 certificate abuse, DCSync/Golden Ticket, AzureHound/ROADrecon cloud enumeration, Microsoft Graph API abuse) to achieve full Active Directory and cloud compromise.
Key facts
| Identifiers | TL-2026-2161 |
|---|---|
| Also known as | A Tale of Two SOCs, AA26-237A |
| Severity | HIGH |
| Affected products | Microsoft Active Directory Certificate Services (AD CS); Microsoft Entra ID / Azure AD (incl. Azure AD Connect); Microsoft System Center Configuration Manager (SCCM) |
| Weaknesses (CWE) | CWE-798, CWE-522, CWE-269, CWE-613 |
| Malware | Rubeus |
| Tools | ADConnectDump, AzureHound, BloodHound - S0521, ROADrecon, Rubeus - S1071 |
| MITRE ATT&CK techniques | T1003 OS Credential Dumping, T1003.006 OS Credential Dumping: DCSync, T1018 Remote System Discovery, T1021.001 Remote Services: Remote Desktop Protocol, T1069.002 Permission Groups Discovery: Domain Groups, T1078 Valid Accounts, T1087.002 Account Discovery: Domain Account, T1114.002 Email Collection: Remote Email Collection, T1136.002 Create Account: Domain Account, T1204.001 User Execution: Malicious Link, T1526 Cloud Service Discovery, T1550.001 Use Alternate Authentication Material: Application Access Token and 7 more in the attack flow |
| Motivation | UNKNOWN |
| Target sectors | government facilities, water and wastewater systems, critical infrastructure |
| Target regions | united states of america |
| Detection rules | 9 (1 shown free) |
| Sources | 7 cited sources |
Detection coverage
As of 2026-08-26, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-2161 across Splunk SPL, Microsoft KQL and Sigma, covering 10 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.
Free detection: ADCS Enrollment-Rights Recon Through ESC1 Certificate Issuance Impersonating a Privileged Account
Detects the full ESC1 kill chain CISA AA26-237A used against Organization A: (1) enumeration of certificate-template enrollment rights and privileged-account names via certutil/Certify/Certipy-style tooling, followed by (2) issuance of a client-authentication certificate from a template with CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT enabled where the resulting certificate's Subject Alternative Name / UPN does not match the requesting account. Either stage alone is a strong signal; both together confirm the impersonation-certificate abuse path.
- Languages: Splunk SPL, Microsoft KQL, Sigma
- ATT&CK: T1649, T1069.002, T1087.002
- Data sources: windows_security, certification_authority_event_log, sysmon, endpoint_command_line_logging
- Severity: critical
- Confidence: high
Splunk SPL
index=windows sourcetype="WinEventLog:Security" (EventCode=4886 OR EventCode=4887 OR EventCode=4888)
| rename Requester_Name as requester, Certificate_Template as template, Subject_Alternative_Name as san
| lookup adcs_client_auth_templates template OUTPUT enrollee_supplies_subject
``` flag templates known/observed to carry CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT + client-auth EKU; maintain the lookup from certutil -v -dstemplate ```
| where enrollee_supplies_subject="true"
| eval san_user=replace(san, "^UPN=", "")
| where isnotnull(san_user) AND NOT (san_user=requester)
| eval stage="issuance_mismatch"
| append
[ search index=edr sourcetype="sysmon" EventCode=1
(Image="*\\certutil.exe" OR Image="*\\certify.exe" OR Image="*\\certipy*" OR Image="*\\certipy.exe")
(CommandLine="*find*" OR CommandLine="*-vulnerable*" OR CommandLine="*-dstemplate*")
| eval requester=User, san_user="(recon-only, no cert yet)", template="(template/enrollment-rights enumeration)", stage="template_recon" ]
| stats count min(_time) as first_seen max(_time) as last_seen values(Certificate_Authority) as ca values(stage) as stages_observed by requester, template, san_user
| where count>=1
| sort -last_seenMicrosoft KQL
let VulnerableTemplates = datatable(TemplateName:string)[
"WebServerAuth", "UserClientAuth", "LegacyClientAuth"
]; // maintain from a periodic Get-ADCSTemplate / certutil audit
let CertRecon = DeviceProcessEvents
| where Timestamp > ago(24h)
| where FileName in~ ("certutil.exe", "Certify.exe", "certipy.exe")
| where ProcessCommandLine has_any ("find", "-vulnerable", "-dstemplate", "template")
| project TimeGenerated = Timestamp, Requester = AccountName, Template = "(enumeration)", SAN = "(recon-only)", Computer = DeviceName, Stage = "template_recon";
let CertIssuance = SecurityEvent
| where TimeGenerated > ago(24h)
| where EventID in (4886, 4887, 4888)
| extend Requester = tostring(EventData.RequesterName), Template = tostring(EventData.CertificateTemplate), SAN = tostring(EventData.SubjectAltName)
| where Template in (VulnerableTemplates)
| where isnotempty(SAN) and SAN !contains Requester
| project TimeGenerated, Requester, Template, SAN, Computer, Stage = "issuance_mismatch";
CertRecon
| union CertIssuance
| summarize StagesObserved = make_set(Stage), count(), min(TimeGenerated), max(TimeGenerated) by Requester, SAN, ComputerSigma
title: ADCS Enrollment-Rights Recon or ESC1 Certificate Request With SAN/Requester Mismatch
id: 7a1e3c2f-4b6d-4e91-9a5c-1f2d8b6e4a01
status: experimental
description: Detects certificate-template/enrollment-rights enumeration (Certify/Certipy/certutil) and/or a certificate issued from a vulnerable AD CS template (CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT + client-auth EKU) whose Subject Alternative Name does not match the requesting account, indicating ESC1 recon and/or privilege-escalation abuse.
references:
- https://www.cisa.gov/news-events/cybersecurity-advisories/aa26-237a
- https://attack.mitre.org/techniques/T1649/
- https://attack.mitre.org/techniques/T1069/002/
- https://attack.mitre.org/techniques/T1087/002/
author: Threadlinqs Team
date: 2026-08-27
tags:
- attack.credential-access
- attack.discovery
- attack.privilege-escalation
- attack.t1649
- attack.t1069.002
- attack.t1087.002
- attack.tool.certify
- attack.tool.certipy
logsource:
product: windows
service: certificationauthority
detection:
selection_issuance:
EventID:
- 4886
- 4887
- 4888
filter_vulnerable_template:
CertificateTemplate|contains:
- 'ClientAuth'
- 'WebServerAuth'
- 'UserAuth'
selection_recon:
Image|endswith:
- '\certutil.exe'
- '\Certify.exe'
- '\certipy.exe'
CommandLine|contains:
- 'find'
- '-vulnerable'
- '-dstemplate'
condition: (selection_issuance and filter_vulnerable_template) or selection_recon
falsepositives:
- Legitimate enrollment agents requesting certificates on behalf of another user via an approved workflow
- Smart-card enrollment stations issuing certificates for helpdesk-assisted onboarding
- PKI administrators or the security team running scheduled Certify/Certipy audits as part of the CISA-recommended template-hardening review
level: highKnown false positives:
- Approved enrollment-agent workflows issuing certificates on behalf of another principal (helpdesk/smart-card onboarding) — exclude by known enrollment-agent SIDs
- PKI administrators running scheduled Certify/Certipy enrollment-rights audits as part of remediation review — allowlist the audit service account
Tuning notes: Maintain the vulnerable-template lookup/datatable from a recurring certutil -v -dstemplate or Certify/Certipy audit of the PKI; alert volume stays near zero once CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT is disabled per CISA's remediation, so any hit post-hardening should page immediately. Chain the recon stage with the issuance-mismatch stage (same requester within a short window) to separate a genuinely malicious enrollment-rights sweep from routine PKI hygiene scans.
- Trusted-Sender Phishing Payload Executes Child Process on Workstation · unlock with TL Creds
- Golden Ticket Indicators — DCSync Precursor, Anomalous Kerberos TGT Encryption, and Seamless-SSO Azure Redemption · unlock with TL Creds
- Seamless SSO Kerberos-to-Azure Authentication Without Prior Interactive Sign-In · unlock with TL Creds
- Unauthorized Machine Account Creation via Default Machine Account Quota, Followed by Its Use and Network Fan-Out · unlock with TL Creds
- Cleartext Credential Access Across SCCM, Application Config, and Saved RDP/FTP Connection Files · unlock with TL Creds
- Disabled Legacy MSOL_ Sync Account Re-Enabled and Granted a New Client Secret · unlock with TL Creds
- DCSync Directory Replication Request From Non-Domain-Controller Host · unlock with TL Creds
- AzureHound / ROADrecon Cloud Enumeration via Distinctive Graph API User-Agent · unlock with TL Creds
Indicators of compromise
As of 2026-08-26, Threadlinqs Intelligence tracks 10 indicators of compromise for TL-2026-2161.
Technical analysis
One SOC never detected the intrusion amid thousands of higher-severity false positives; the other contained the initial phishing within 2-20 minutes and later caught the cloud-enumeration phase via Entra ID risky-user alerts on the AzureHound user agent and anomalous Graph API volume.
On 2026-08-25, CISA published AA26-237A, comparing two concurrent red team assessments run with the same tradecraft against critical infrastructure organizations to illustrate how detection maturity — not tooling — determines outcome.
Against Organization A (Government Services and Facilities Sector), the red team found a web application still using default credentials for built-in accounts, used that foothold to send phishing emails from a trusted internal address, and landed payloads on four workstations. From there it abused the domain's default Machine Account Quota (MAQ) to add unauthorized machine accounts, then exploited an ESC1-class misconfiguration in Active Directory Certificate Services — templates with `CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT` enabled and client-authentication EKUs — to request certificates impersonating privileged accounts. The team recovered cleartext database and service credentials from configuration files (including a `connections.json`-style file) and long-lived, non-expiring static AWS IAM access keys, then pivoted to the cloud by harvesting a Primary Refresh Token and abusing over-permissioned Entra ID applications, ultimately reading mail across the tenant. The SOC never detected any of it: staff were buried under thousands of false-positive alerts (many scored higher-severity than the actual intrusion), multiple siloed SOCs ran different EDR stacks with no shared visibility, and there were no written escalation procedures. The red team confirmed the miss by reading the defenders' own SOC email.
Against Organization B (Water and Wastewater Systems Sector), the same phishing payload was detected on execution and the affected workstations were isolated within 2-20 minutes, cutting off C2 before the intrusion could spread — so the engagement shifted to an assume-breach model from a designated non-privileged host. From there the team found a domain service account's password stored in cleartext inside an XML file on an SCCM distribution point; that account held rights sufficient to perform a DCSync attack and obtain the `krbtgt` hash. The team forged a Golden Ticket and used the resulting Kerberos tickets — via Seamless SSO — to authenticate to Azure without any user's cleartext password, running AzureHound, ROADrecon, and ADConnectDump for cloud enumeration and abusing Microsoft Graph API scopes (Mail.Read, Mail.ReadWrite, Chat.Read.All, Files.Read.All, Application.ReadWrite.All, AppRoleAssignment.ReadWrite.All) — at one point re-enabling a disabled, MFA-exempt legacy AD-sync (MSOL_) account, DCSyncing its credentials, and adding a client secret to gain full mailbox access across the organization. The team also reached an OT-network bastion host reached via FTP credentials on a jump server, but outbound access was blocked and no C2 channel was established there. This time Organization B's custom Entra ID risky-user detections — "Unfamiliar sign-in properties" and "Suspicious API traffic" — flagged the AzureHound HTTP user agent and accounts exceeding Graph API request thresholds, and the SOC triaged, coordinated with engineering, and reimaged affected hosts.
Neither organization had Conditional Access enabled for workload identities, letting applications with broad Graph permissions bypass normal Conditional Access rules in both cases. CISA's core recommendation is architectural and procedural: disable `CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT` and restrict enrollment on ADCS templates, zero out Machine Account Quota where unneeded, eliminate cleartext credentials and rotate long-lived cloud keys, enforce Conditional Access for workload identities, treat SCCM as a Tier 0 asset, and — the advisory's central point — tune alerting to cut false-positive volume and give SOC staff clear escalation authority, since Organization B's success came from process maturity rather than superior tools.
MITRE ATT&CK techniques by tactic:
Credential Access
T1003 OS Credential Dumping; T1003.006 OS Credential Dumping: DCSync; T1552.001 Unsecured Credentials: Credentials In Files; T1558.001 Steal or Forge Kerberos Tickets: Golden Ticket; T1649 Steal or Forge Authentication Certificates
Discovery
T1018 Remote System Discovery; T1069.002 Permission Groups Discovery: Domain Groups; T1087.002 Account Discovery: Domain Account; T1526 Cloud Service Discovery; T1615 Group Policy Discovery
Lateral Movement
T1021.001 Remote Services: Remote Desktop Protocol
Initial Access
T1078 Valid Accounts; T1566 Phishing
Collection
T1114.002 Email Collection: Remote Email Collection
Persistence
T1136.002 Create Account: Domain Account
Execution
T1204.001 User Execution: Malicious Link
lateral-movement
T1550.001 Use Alternate Authentication Material: Application Access Token
Reconnaissance
T1589.001 Gather Victim Identity Information: Credentials; T1589.002 Gather Victim Identity Information: Email Addresses
Remediation
Immediate actions
- Disable CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT on ADCS certificate templates and restrict who can enroll for client-authentication EKUs
- Set Machine Account Quota (MAQ) to zero for accounts that do not operationally need to join machines to the domain
- Remove cleartext service/database credentials from SCCM distribution-point files, configuration files, and network shares
- Rotate and revoke long-lived static AWS IAM access keys; require expiration
- Disable or remove unused legacy AD-sync (MSOL_) accounts, or enforce MFA on them
Workarounds
- Enable Conditional Access policies for workload identities (service principals/applications), not just user identities
- Audit Entra ID application permissions and remove excessive Microsoft Graph scopes (Mail.ReadWrite, Application.ReadWrite.All, AppRoleAssignment.ReadWrite.All)
- Monitor for the AzureHound HTTP user-agent string and for accounts exceeding normal Microsoft Graph API request-volume thresholds
Longer-term hardening
- Consolidate SOC tooling and establish shared visibility across previously siloed SOCs/EDR stacks
- Implement alert-baseline tuning to reduce false-positive volume so real intrusions are not buried
- Write and rehearse incident escalation procedures; give SOC analysts authority to contain hosts without bureaucratic delay
- Treat SCCM (and similar endpoint-management platforms) as a Tier 0 asset with domain-controller-level protection
- Establish a token/credential revocation procedure for compromised PRTs and OAuth application secrets
Affected and fixed versions
- Microsoft — Active Directory Certificate Services (AD CS)
Vulnerable versions: templates with CT_FLAG_ENROLLEE_SUPPLIES_SUBJECT + client-auth EKU (ESC1) misconfigured
Fixed in: templates with enrollee-supplies-subject disabled and enrollment restricted to intended principals - Microsoft — Entra ID / Azure AD (incl. Azure AD Connect)
Vulnerable versions: tenants without Conditional Access for workload identities; over-permissioned applications; legacy MSOL_ sync accounts without MFA
Fixed in: N/A - configuration hardening, not a version fix - Microsoft — System Center Configuration Manager (SCCM)
Vulnerable versions: distribution points storing cleartext service-account credentials in configuration/XML files
Fixed in: N/A - configuration hardening, not a version fix
Timeline
- Organization B's Entra ID risky-user detections flag the AzureHound user agent and anomalous Microsoft Graph API request volume; the SOC triages, coordinates with engineering, and reimages affected hosts before the OT-adjacent bastion host can be leveraged further.
- Red team forges a Golden Ticket and uses Kerberos-based Seamless SSO to authenticate to Azure, running AzureHound, ROADrecon, and ADConnectDump for cloud enumeration and abusing Microsoft Graph API scopes to access mailboxes tenant-wide.
- From an assume-breach foothold, the red team finds a domain service account's password stored in cleartext inside an XML file on an SCCM distribution point and performs a DCSync attack to obtain the krbtgt hash.
- The same phishing payload against Organization B (Water and Wastewater Systems) is detected on execution; the SOC isolates affected workstations within 2-20 minutes, forcing the red team into an assume-breach model.
- Red team recovers cleartext credentials and non-expiring static AWS IAM keys, harvests a Primary Refresh Token, and abuses over-permissioned Entra ID applications to read tenant mail; Organization A's SOC never detects the intrusion amid thousands of higher-severity false positives.
- Red team abuses Organization A's default Machine Account Quota to add unauthorized machine accounts and exploits an ESC1-vulnerable ADCS certificate template to request certificates impersonating privileged accounts.
- Red team gains initial access to Organization A (Government Services and Facilities) via a web application using default credentials, then sends phishing emails from a trusted internal address, landing payloads on four workstations.
- CISA publishes AA26-237A, "A Tale of Two SOCs," comparing two concurrent red team assessments against critical infrastructure organizations.
- SecurityAffairs, The Hacker News, and other outlets report on AA26-237A's findings as a case study in SOC detection maturity across critical infrastructure sectors.
Sources cited
- A Tale of Two SOCs: Insights From Two Red Team Assessments (AA26-237A) CISA · Government advisory · 2026-08-25
- CISA Advisory Highlights Red Team Findings to Help Organizations Assess Risk, Identify Threats and Enable Effective Incident Response CISA · Government advisory · 2026-08-25
- Cybersecurity Advisory: A Tale of Two SOCs (Vulnerability Vault) community.opentextcybersecurity.com · Blog · 2026-08-25
- A Tale of Two SOCs: Insights From Two Red Team Assessments assurantcyber.com · Blog · 2026-08-25
- A Tale of Two SOCs: Insights From Two Red Team Assessments (mirror) hendryadrian.com · Blog · 2026-08-25
- CISA Red Team Fully Compromised Two Critical Infrastructure Orgs securityaffairs.com · News · 2026-08-26
- CISA Red Team Compromised Two Critical Infrastructure Orgs, One Detected Nothing The Hacker News · News · 2026-08-26
Previous and next in vulnerability
- Apache Tomcat 11.0.25 Fixes 11 Vulnerabilities Including HTTP/2 DoS, Authorization Bypass, and Auth Fail-Open Flaws — NVD Scores 5 of 11 CRITICAL/HIGH Despite Apache's Low/Moderate Ratings Previous report
- Critical WatchGuard Agent for Windows Flaws (CVE-2026-57910, CVE-2026-57909) Enable Unauthenticated SYSTEM-Level RCE Next report
Drafted by the Threadlinqs research pipeline from 7 sources; published ; last updated —.