CVE-2026-77937 in Letta
Summary
by MITRE • 10/20/2026
Letta through 0.16.8 contains an authentication bypass vulnerability that allows API callers to impersonate other users by supplying a self-asserted user_id HTTP header. Attackers can set user_id to another tenant's ID, even with LETTA_SERVER_SECURE enabled, to read and modify that user's agents, memory and configuration.
Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.
Analysis
by VulDB Data Team • 10/20/2026
The vulnerability identified in Letta versions through 0.16.8 represents a critical authentication bypass flaw rooted in the improper validation of identity assertions within the application’s API layer. This security defect allows unauthorized actors to impersonate legitimate users by manipulating specific HTTP headers, specifically the user_id field. The core technical issue lies in the server's reliance on client-supplied data for session management and access control decisions without sufficient verification against a trusted authority or secure backend state. By accepting a self-asserted user ID, the application fails to enforce proper identity proofing mechanisms that are standard in robust authentication architectures. This flaw effectively neutralizes the intended security boundaries between different tenants or users within the multi-user environment of the Letta platform.
From an operational perspective, this vulnerability enables severe unauthorized access and data manipulation. An attacker can simply inject a target user's identifier into the HTTP request header to bypass any existing login requirements or session checks. This capability allows the adversary to read sensitive information belonging to other users, including their AI agents' configurations, stored memory states, and personal settings. Furthermore, the ability to modify these elements means an attacker could alter agent behaviors, inject malicious instructions into a user's context window, or delete critical data associated with another tenant. The impact is particularly acute because it persists even when the LETTA_SERVER_SECURE environment variable is enabled, indicating that basic security flags are insufficient to mitigate this specific logic flaw in identity handling.
This vulnerability aligns closely with CWE-287, which describes Improper Authentication, as well as CWE-613, Insufficient Session Expiration, though it specifically manifests through header manipulation typical of CWE-915, Improper Modification of Dynamically-Determined Object Attributes. In the context of the MITRE ATT&CK framework, this behavior corresponds to T1078, Valid Accounts, where an attacker uses legitimate credentials or identity tokens obtained from another source to gain access. It also relates to T1496, Resource Hijacking, if the impersonated user's resources are abused for computational purposes. The attack vector is classified as Network-based with Low complexity, requiring no prior authentication and minimal technical skill to exploit using standard HTTP tools.
To mitigate this vulnerability, immediate action must be taken by upgrading Letta to a version later than 0.16.8 where the identity verification logic has been corrected. In the interim, administrators should implement strict input validation on all API endpoints that process user identification headers. It is critical to ensure that server-side session management relies exclusively on cryptographically secure tokens issued by an authenticated authentication service rather than trusting client-supplied identifiers. Additionally, enabling comprehensive logging and monitoring for anomalous header values can help detect exploitation attempts in real-time. Organizations should also enforce strict CORS policies and consider implementing mutual TLS or API key validation to add layers of defense-in-depth against identity spoofing attacks.