
A conceptual tour of how Kerberos tickets are abused in Active Directory environments, written for defenders and authorized security teams studying detection theory.
| Topic | Kerberos ticket abuse concepts in Active Directory environments |
| Category | Authentication protocol analysis |
| Primary Use | Understanding ticket-based abuse patterns to build detection and hardening strategies |
| Safe Use | Educational, defensive analysis for authorized security teams operating in labs or environments they are contracted to assess |
| Telemetry Note | Kerberos abuse typically leaves traces in KDC event logs, ticket sizes, encryption type downgrades, and anomalous TGT/ST request patterns observable by defenders |
Kerberos has been the default authentication workhorse of Active Directory for decades, and its longevity is precisely why it occupies such a central place in offensive tradecraft discussions. The protocol itself is elegant: a client authenticates once to a Key Distribution Center (KDC), receives a Ticket Granting Ticket (TGT), and then exchanges that TGT for Service Tickets (ST) as needed to access resources. Every ticket is a structured, encrypted blob of identity assertions. The interesting analytical question for defenders is not how to break the cryptography — modern Kerberos encryption types like AES256 remain robust — but what happens when the surrounding design assumptions are violated by legitimate protocol features used in illegitimate ways.
The core insight worth internalizing is that most Kerberos abuse does not exploit a bug in the protocol. It exploits the gap between what a ticket cryptographically guarantees and what the environment assumes it guarantees. A ticket proves the bearer authenticated at some point with credentials that mapped to an account. It says nothing about whether the human currently presenting that ticket is the same human, whether the account should have access to that particular service, or whether the ticket's contents — group memberships, privilege attributes, validity lifetime — match the directory's current intent. Attackers who understand this gap can operate entirely inside the protocol, generating traffic that a naive firewall or even a compliant KDC considers perfectly valid.
Consider the canonical example of a privilege attribute forgery technique, commonly known by its tooling name family built around requests for tickets with inflated PAC contents. In this pattern, an attacker with control of an account that possesses a particular directory-level delegation — the ability to modify some other principal's attributes — leverages that write access to request a ticket that embeds elevated group memberships. The KDC, following its own rules, faithfully signs the modified data. From the defender's perspective, the alarming property is that the resulting ticket is cryptographically authentic. No signature was broken, no key was stolen. The abuse lives in the semantic layer, and detecting it requires understanding what the directory's delegation model actually permitted, which is a configuration audit problem rather than a packet inspection problem.
A second conceptual family revolves around ticket persistence. Because a TGT is a self-contained bearer credential with a lifetime set by domain policy, possession of the ticket is functionally equivalent to possession of the password for that lifetime — and renewable tickets can extend that window substantially, particularly in environments where account-based revocation is weak. Older RC4-encrypted tickets associated with accounts whose krbtgt keys have never been rotated present an even longer memory: forged tickets minted from a compromised krbtgt key remain usable until that key changes twice. The defensive lesson is that Kerberos introduces a class of state that outlives credential resets, and incident response plans that only reset user passwords without considering ticket lifetimes, krbtgt rotation, and lingering cached tickets on endpoints are structurally incomplete.
Constrained and unconstrained delegation form a third rich area of conceptual analysis. Delegation is a legitimate feature: it lets a service act on behalf of a user when accessing a backend resource. The abuse surface emerges when the protocol's trust relationships — a service trusted for delegation, a protocol transition permitted, a resource-based delegation configured via msDS-AllowedToActOnBehalfOfOtherIdentity — are combined with attacker-controlled accounts or services. The classic pattern involves an attacker creating or compromising a machine account with delegation rights and then requesting tickets that impersonate a privileged user toward a target service. What makes this family analytically interesting is that every step is a legitimate directory operation; the attack is an emergent property of configuration composition, which is why automated attack-path tooling has made delegation misconfiguration such a prominent finding in modern Active Directory security reviews.
Encryption type downgrades deserve their own paragraph because they illustrate how historical compatibility creates modern risk surfaces. Environments that still permit RC4-HMAC alongside AES inherit the weaker key derivation of the older suite, where keys derive trivially from NT-style password hashes. Requesting tickets encrypted with legacy types can therefore turn a weak password into a crackable offline artifact, entirely outside the view of the KDC's audit trail once the ticket is issued. The defensive countermeasure is well understood — enforce AES-only via domain functional level and group policy — yet the telemetry angle matters too: a sudden prevalence of RC4-encrypted ticket requests in an environment that has migrated to AES is itself a high-signal indicator worth alerting on.
Another concept with defensive relevance is the abuse of ticket request volume and naming patterns. Offensive tooling historically made assumptions that leaked into observable behavior: machine account naming conventions, predictable service principal name requests, anomalous rates of TGT requests from a single host, or ST requests for services the host has never talked to before. Windows auditing surfaces much of this through event classes covering Kerberos authentication and ticket operations, particularly when advanced audit policies for account logon and Kerberos service ticket events are enabled. The analytical point is that detection engineering for Kerberos abuse is fundamentally about baselines: what services does each host normally request, what encryption types does it normally use, and how long are the tickets it normally obtains?
It is also worth separating concept from capability when reading public research in this space. Many well-known techniques are named after the tools that popularized them rather than the underlying protocol behavior, which can obscure the fact that the underlying primitives — ticket requests, delegation attributes, key version numbers, PAC structures — are all documented protocol elements. For a defender, mapping each publicized technique back to the directory attribute or protocol field it manipulates is the most productive exercise, because that mapping generalizes: new tooling will keep appearing, but the field it touches remains stable, and so should the telemetry strategy. This is also why configuration hardening — least-privilege delegation, Protected Users group membership for sensitive accounts, krbtgt rotation discipline, removal of legacy encryption — consistently outperforms signature-based detection over the long term.
Finally, the ethical framing matters as much as the technical one. Everything discussed here is documented in RFCs, Microsoft's own protocol documentation, and years of peer-reviewed conference research. Understanding these concepts is essential for authorized penetration testers who need to explain risk to their clients, for detection engineers building Kerberos telemetry, and for architects designing directory delegation models that fail closed. The same knowledge used to harden an estate is the knowledge that would be needed to attack it; in mature security practice, the two are the same literacy. Any experimentation with these techniques belongs strictly in lab environments or engagements governed by explicit written authorization, and the goal of analysis should always be the defensive artifact: a hardened configuration, a detection rule, or a documented attack path eliminated.
Educational analysis for authorized security professionals. Use only in controlled, authorized environments.
Related coverage
0 comentários:
Post a Comment
Note: Only a member of this blog may post a comment.