In short
DCSync is an attack technique that abuses Active Directory replication. An attacker who holds replication rights pretends to be a domain controller and asks a real one to send account data, including password hashes such as the KRBTGT hash. No code runs on the domain controller, so the attack leaves few traces.
Key takeaways
- What it is
- A credential theft technique, MITRE ATT&CK T1003.006, that uses the directory replication protocol to request password data from a domain controller.
- How it works
- A tool such as Mimikatz or Impacket calls the replication interface (MS-DRSR) from a normal computer, using an account that has replication rights.
- Why it matters
- The KRBTGT and administrator hashes it returns let an attacker forge Kerberos tickets and control the whole domain, often for a long time.
- What Odin Research saw
- An exposed staging server in Monograph 01 held DCSync stager scripts built on Impacket, part of a planned path from a web shell to the domain controller.
- What to do first
- Audit who holds Replicating Directory Changes rights, enable replication auditing (Event 4662), and alert on replication from any machine that is not a domain controller.
What is DCSync?
DCSync is an attack technique that abuses Active Directory replication to steal password data. An attacker who controls an account with replication rights pretends to be a domain controller and asks a real domain controller to replicate account secrets. The response includes password hashes, among them the hash of the KRBTGT account.
MITRE ATT&CK lists DCSync as sub-technique T1003.006 of OS Credential Dumping. MITRE says the recovered hashes, including KRBTGT and administrator hashes, can be used to forge Golden Tickets, pass tickets or change account passwords.
The attack needs no malware on the domain controller and no access to its disk. It uses a normal, documented protocol. That is what makes it attractive: it looks like replication, and replication happens all the time in a healthy Active Directory.
How a DCSync attack works
Domain controllers keep each other in sync with the Directory Replication Service Remote Protocol, MS-DRSR. Microsoft documents its IDL_DRSGetNCChanges method as the call that replicates updates from a naming context replica on the server.
- Get the right account. The attacker needs an identity with the "Replicating Directory Changes" and "Replicating Directory Changes All" permissions. MITRE notes that members of Administrators, Domain Admins or Enterprise Admins, and domain controller computer accounts, can run DCSync.
- Call the replication interface. From any machine that can reach a domain controller, a tool binds to MS-DRSR and requests changes for one account or for all of them.
- Receive the secrets. The domain controller answers as it would to a peer, with password hashes and other secret attributes.
- Use them. The attacker cracks the hashes, passes them, or forges Kerberos tickets with the KRBTGT hash.
MITRE names Mimikatz and its lsadump module as the main implementation. Impacket offers the same capability in Python, which is what Odin Research found staged in Monograph 01.
How attackers use DCSync
DCSync is a late-stage step. Attackers use it once they hold a privileged account, to collect every credential in the domain at once. MITRE ATT&CK lists groups that have used it, including APT29 during the SolarWinds campaign, LAPSUS$, Mustang Panda, Scattered Spider and Storm-0501.
AI tools have entered the picture too. MITRE's entry for campaign C0064 says an adversary had Claude Code use DCSync to pull Active Directory password data during attacks on Mexican government bodies.
With domain-wide hashes in hand, an attacker can move to any server that trusts the domain. See lateral movement.
What Odin Research observed
Monograph 01 describes an exposed staging server with 364 attack files. They stage a path from a web shell through SQL Server and Oracle databases to Active Directory replication and VMware ESXi probing. Odin holds no logs from the targeted network, so every step is staged capability unless a captured output says otherwise.
- A dedicated DCSync stage. The operator kept a numbered folder named for DCSync and the domain controller, and a batch file that chains tunnel setup, SQL escalation and DCSync.
- Impacket-based stagers. A 3,452-byte
dcsync.pycalls Impacket'sNTDSHashes(IDL_DRSGetNCChanges) through a local SOCKS5 proxy. A shorter draft has an indentation bug. Other scripts test access to the domain controller'sC$share and build raw LDAP bind packets by hand. - A temporary Domain Admin. One script creates a temporary domain account and adds it to Domain Admins for replication. A rollback script deletes it later.
- Not verified. No output of replicated hashes was captured, so hash replication is not shown.
- Few traces by design. The report notes that DCSync and raw LDAP operations write no files to the domain controller. Only event logs and network records keep a trace.
The report weighs a hostile intrusion against an authorized red team and a lab. Read the short read for a summary.
How to detect DCSync
- Replication from non-DCs. Microsoft Defender for Identity raises "Suspected DCSync attack (replication of directory services)" when a replication request comes from a computer that is not a domain controller. Install its sensor on every domain controller.
- Event 4662. With Audit Directory Service Access enabled and a suitable SACL on the domain object, domain controllers log Event 4662 for operations on directory objects. Alert on Control Access (
0x100) with a replication extended-right GUID from an account that is not a domain controller or your directory sync account. - Exact allowlists. List domain controller machine accounts and sync service accounts by exact name. Wildcards such as
*$let an attacker hide behind a matching name. - Network view. MS-DRSR traffic between a workstation or server and a domain controller is unusual.
- Short-lived admins. Accounts created, added to Domain Admins and deleted within days often surround a DCSync.
Odin Research publishes rules for these. Monograph 01 carries a Sigma rule for replication-right access from non-allowlisted principals, a Sentinel KQL query for short-lived admin accounts and Suricata signatures for cleartext LDAP simple binds. All are in the Defense Kit.
How to defend against DCSync
- Review replication rights. MITRE recommends managing the access control list for "Replicating Directory Changes" and other replication permissions (M1015). Microsoft advises validating who holds Replicate Directory Changes and Replicate Directory Changes All.
- Tier privileged accounts. Keep domain admins off workstations and member servers, and avoid domain accounts in local administrator groups (M1026). See least privilege.
- Use unique local admin passwords across systems (M1027), so one stolen hash does not unlock every server.
- Respond fully. If DCSync succeeded, reset the KRBTGT password twice, reset privileged accounts, and look for forged tickets. Related Kerberos abuse is covered in Kerberoasting.
DCSync compared with other credential dumping methods
| Method | Where it runs | What it needs | Main trace |
|---|---|---|---|
| DCSync | Any host that reaches a DC | Replication rights | Event 4662, replication from a non-DC |
| LSASS memory dump | The target host | Local admin or SYSTEM | Process access to lsass.exe |
| NTDS.dit copy | The domain controller | Admin access on a DC | Shadow copies and file access on the DC |
| Kerberoasting | Any domain host | Any domain account | Many service ticket requests |
Frequently asked questions about DCSync
What is DCSync?
DCSync is an attack technique in which an account with replication rights impersonates a domain controller and asks a real one for password hashes through the Active Directory replication protocol.
What permissions does DCSync need?
The Replicating Directory Changes and Replicating Directory Changes All rights. MITRE notes that members of Administrators, Domain Admins and Enterprise Admins, and domain controller computer accounts, can run DCSync.
What is the MITRE ATT&CK ID for DCSync?
T1003.006, a sub-technique of OS Credential Dumping (T1003) in the Credential Access tactic.
Which tools perform DCSync?
Mimikatz, through its lsadump module, is the best-known tool. Impacket offers the same capability in Python, as seen in the scripts Odin Research described in Monograph 01.
How do I detect a DCSync attack?
Alert on replication requests from machines that are not domain controllers, and on Event 4662 with replication rights from unexpected accounts. Microsoft Defender for Identity has a built-in alert, and Odin Research publishes a Sigma rule.
Does DCSync leave logs?
It writes no files on the domain controller. With directory service auditing and a replication SACL, domain controllers log Event 4662, and network sensors can see the replication traffic.
How do I stop DCSync?
Limit replication rights to domain controllers and sync accounts, protect privileged accounts with tiering, and monitor replication. If it succeeded, reset the KRBTGT password twice.
DCSync vs Kerberoasting: what is the difference?
DCSync needs replication rights and returns password hashes for any account directly from a domain controller. Kerberoasting needs only a normal domain account and yields service tickets that must be cracked offline.
Sources
- MITRE ATT&CK, T1003.006 OS Credential Dumping: DCSync. attack.mitre.org/techniques/T1003/006/ (checked 21 November 2026)
- MITRE ATT&CK, Campaign C0064: AI-Assisted Breach of Mexico's Government Infrastructure. attack.mitre.org/campaigns/C0064/ (checked 21 November 2026)
- Microsoft Open Specifications, [MS-DRSR] IDL_DRSGetNCChanges. learn.microsoft.com/en-us/openspecs/windows_protocols/ms-drsr/b63730ac-614c-431c-9501-28d6aca91894 (checked 21 November 2026)
- Microsoft Learn, Microsoft Defender for Identity classic security alerts. learn.microsoft.com/en-us/defender-for-identity/alerts-mdi-classic (checked 21 November 2026)
- Microsoft Learn, Event 4662: An operation was performed on an object. learn.microsoft.com/en-us/previous-versions/windows/it-pro/windows-10/security/threat-protection/auditing/event-4662 (checked 21 November 2026)
Related tools and terms
What is OAST?
Out-of-band application security testing (OAST) detects blind vulnerabilities by making a target call back to a server the tester controls, usually over DNS or HTTP.
TechniqueWhat is JNDI injection?
JNDI injection makes a Java application look up a remote object, usually over LDAP or RMI, and load attacker code. Log4Shell is the best-known example.
TechniqueWhat is HVNC?
Hidden VNC (HVNC) creates a second, invisible desktop on the victim's computer so an attacker can use the victim's browser sessions unseen.
VulnerabilityWhat is Ghostcat?
Ghostcat (CVE-2020-1938) is a flaw in the Apache Tomcat AJP connector that lets an attacker read files and, in some setups, run code.
TechniqueWhat is JARM?
JARM is an open TLS server fingerprinting method from Salesforce. It hashes how a server answers a set of crafted TLS handshakes.
TechniqueWhat is xp_cmdshell?
xp_cmdshell is a SQL Server extended stored procedure that runs operating system commands. Attackers enable it to move from database access to the host.
Odin Research
Get the rules, not just the definition
Every Odin Research report ships with verified indicators and detection rules you can load today.