CVE-2026-85105 in hermes-agent
Summary
by MITRE • 09/03/2026
A flaw has been found in NousResearch hermes-agent 0.18.0. Affected by this issue is the function _sess_nowait of the file s71.py of the component Session Management. This manipulation of the argument session_id causes authorization bypass. The attack can be initiated remotely. The vendor was contacted early about this disclosure but did not respond in any way.
Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.
Analysis
by VulDB Data Team • 09/03/2026
The vulnerability identified within NousResearch hermes-agent version 0.18.0 represents a critical security flaw located specifically within the Session Management component, affecting the _sess_nowait function found in the s71.py module. This issue stems from an improper manipulation of the session_id argument passed to this function. In secure software architecture, session identifiers serve as the primary mechanism for maintaining state and authenticating user interactions across network requests. When a system fails to properly validate or sanitize these identifiers during processing, it creates a pathway for attackers to subvert authentication controls. The specific nature of this flaw allows an adversary to bypass authorization checks by supplying crafted input that exploits the logic within _sess_nowait, effectively allowing unauthorized access to protected resources without valid credentials.
From a technical perspective, this vulnerability aligns with CWE-287, which describes Improper Authentication, and more specifically CWE-345 regarding Insufficient Verification of Data Authenticity or Integrity in session management contexts. The root cause lies in the application's failure to ensure that the session_id provided by the client corresponds to a valid, active, and authorized session on the server side before granting access. By manipulating this argument, an attacker can potentially impersonate other users or escalate privileges within the application environment. This type of flaw is particularly dangerous because it undermines the fundamental trust model upon which web applications are built, allowing malicious actors to operate with the permissions of legitimate users while remaining undetected by standard security monitoring tools that rely on successful authentication events for logging and alerting.
The operational impact of this vulnerability is severe due to its remote exploitability. Since the attack can be initiated remotely over a network connection, such as via HTTP or HTTPS requests, there is no need for physical access or prior local foothold within the target system. This characteristic significantly lowers the barrier to entry for potential attackers, enabling automated scanning tools and opportunistic adversaries to identify and exploit vulnerable instances across the internet with minimal effort. Successful exploitation could lead to a complete compromise of user accounts, unauthorized data exfiltration, modification of critical application state, or further lateral movement within an internal network if hermes-agent is deployed in a broader infrastructure context. The lack of response from NousResearch regarding early disclosure attempts exacerbates the risk, as it leaves users without official patches or guidance for immediate remediation during this window of exposure.
To mitigate the risks associated with this authorization bypass vulnerability, organizations deploying hermes-agent 0.18.0 should implement compensating controls immediately if an update is not yet available. Network-level defenses such as Web Application Firewalls can be configured to detect and block anomalous session ID patterns or requests that exhibit characteristics of injection attacks. Additionally, implementing strict input validation on the server side for all session-related parameters is crucial; developers must ensure that session IDs are validated against a secure store before any processing occurs in functions like _sess_nowait. It is also recommended to enforce short-lived sessions and rotate identifiers after significant state changes to limit the window of opportunity for exploitation. Until an official patch is released by NousResearch, administrators should consider restricting access to hermes-agent interfaces through IP whitelisting or multi-factor authentication layers that operate independently of the vulnerable session management logic to provide a secondary barrier against unauthorized access attempts.