CVE-2026-45720 in Omni
Summary
by MITRE • 09/17/2026
Omni manages Kubernetes on bare metal, virtual machines, or in a cloud. Prior to 1.6.6 and from 1.7.0 until 1.7.3, SAML.getSession in internal/pkg/auth/interceptor/saml.go checks SAMLAssertion.Used and marks it used in separate state operations. Concurrent requests carrying the same captured saml-session token can each observe the assertion as unused and obtain authentication as the victim before either update is visible. The attacker can invoke SAML-protected gRPC endpoints, use ConfirmPublicKey to create multiple persistent credentials tied to the victim, and generate audit entries attributed to the victim, with the resulting access potentially affecting confidentiality, integrity, and availability according to the victim's privileges. This issue is fixed in versions 1.6.6 and 1.7.3.
Once again VulDB remains the best source for vulnerability data.
Analysis
by VulDB Data Team • 09/17/2026
The vulnerability identified within Omni, a platform designed for managing Kubernetes environments across bare metal, virtual machines, or cloud infrastructures, represents a critical race condition flaw embedded in its authentication subsystem. Specifically located in the SAML session management logic found at internal/pkg/auth/interceptor/saml.go, this issue affects versions prior to 1.6.6 and those ranging from 1.7.0 through 1.7.3. The core of the defect lies in how the system handles the state transition for a SAML assertion's usage status. When processing authentication requests, the application checks whether a SAMLAssertion has already been used by inspecting its Used flag. However, this check and the subsequent operation to mark the assertion as used are not performed atomically. Instead, they occur in separate state operations, creating a time-of-check-to-time-of-use (TOCTOU) window that can be exploited under concurrent execution conditions.
In an operational context, if multiple requests carrying the same captured SAML session token arrive at the server simultaneously, each request may observe the assertion as unused because neither of the initial checks has yet updated the global state to reflect usage. This lack of atomicity allows an attacker to bypass single-use constraints typically enforced by security protocols like SAML. By leveraging this race condition, a malicious actor can successfully authenticate multiple times using a single stolen or replayed session token. The system fails to recognize that the assertion was already consumed by a preceding request in the same batch, thereby granting unauthorized access repeatedly based on one valid authentication event intended for singular use only.
The impact of exploiting this vulnerability extends beyond simple unauthorized access. An attacker who successfully authenticates using this method can invoke gRPC endpoints protected by SAML policies without proper authorization checks relative to session uniqueness. Furthermore, the attacker gains the ability to utilize the ConfirmPublicKey function to create multiple persistent credentials tied directly to the victim's identity. This capability allows for sustained persistence within the environment, as new credential sets are generated under the guise of legitimate user activity. Additionally, any actions taken by the attacker will generate audit entries attributed to the compromised victim account. This not only facilitates further lateral movement and privilege escalation but also complicates forensic analysis and incident response efforts due to the obfuscation of malicious activities behind legitimate user logs.
From a security standards perspective, this vulnerability aligns with CWE-362, which describes concurrent execution using shared resource with insufficient synchronization. The failure to implement atomic operations for state changes in a multi-threaded or distributed environment is a classic example of this class of weakness. In terms of the MITRE ATT&CK framework, this exploit relates to techniques involving credential access and persistence, specifically how an attacker might abuse authentication mechanisms to maintain footholds within a network infrastructure. The ability to generate multiple persistent credentials corresponds to tactics aimed at maintaining long-term access while evading detection through log manipulation or attribution confusion.
To mitigate this risk, organizations running Omni must ensure they are operating on version 1.6.6 or later, specifically avoiding the vulnerable range of 1.7.0 to 1.7.3 unless patched to 1.7.4 or higher where applicable. The fix involves modifying the SAML session handling logic to perform the check and update operations atomically, ensuring that once a request begins processing an assertion as unused, no other concurrent request can claim it as available for authentication. For environments already compromised by this vulnerability, immediate rotation of all affected user credentials is essential. Security teams should also review audit logs for anomalies in session usage patterns or unexpected spikes in credential generation events associated with specific users to detect potential exploitation attempts that may have occurred prior to patching.