NousResearch hermes-agent CVE-2026-7112

The API server component must be exposed to network access (remotely reachable by the attacker). The vulnerability exists in the default authentication validation logic of _check_auth, meaning any hermes-agent 0.8.0 deployment with an active API server is potentially vulnerable without additional compensating controls. No special configuration is required to trigger the flaw-the improper authentication logic is inherent to the affected version. The attack does not require user interaction or valid credentials. The high complexity (AC:H) likely stems from the specific crafting requirements for the malicious request (e.g., correct payload format, timing, or header manipulation) rather than environmental factors.

This vulnerability presents moderate real-world risk despite a CVSS score of 6.3. The CVSS vector AV:N (network-accessible) and PR:N (no privileges required) indicate remote unauthenticated exploitation is possible, but AC:H (high complexity) and the description stating 'exploitation appears to be difficult' suggest practical attack barriers exist. EPSS data is not provided, limiting probabilistic exploitation assessment. The publicly available exploit code increases real-world risk by lowering the barrier to attack execution. KEV status is not provided; however, the project's lack of response to early notification combined with public disclosure suggests limited defensive awareness. The confined scope (VC:L/VI:L/VA:L - low confidentiality, integrity, and availability impact) indicates this is not a critical severity vulnerability despite network accessibility. The actual risk depends on whether hermes-agent is deployed in internet-facing environments and whether the API_SERVER_KEY validation is essential to the security posture of downstream systems relying on authenticated API calls.

An attacker crafts a malicious API request to a hermes-agent instance and sends it over the network to the API server endpoint. The _check_auth function fails to properly validate the API_SERVER_KEY due to its logic flaw, allowing the attacker to bypass authentication. The attacker gains unauthorized access to the API and can execute subsequent actions depending on the API's capabilities, such as data exfiltration or system manipulation. With publicly available exploit code, the barrier to execution is significantly lowered.