The Hermes Agent webhook gateway builds a session chat_id using the pattern webhook:{route_name}:{delivery_id}. It then recreates the authorizing route by splitting that string. The flaw, a CWE‑863 vulnerability (improper authorization), occurs when a supplied route name contains a colon. The split results in a higher‑privileged prefix route, bypassing the intended authorization. With a valid target HMAC, an attacker can access the toolsets of that privileged route, creating unauthorized elevation of privilege.

Nous Research:Hermes Agent is the affected product. The CVE description does not specify exact vulnerable versions. The flaw was identified in the code path related to route name parsing before the release of v0.22.2, but without official version impact data it is unclear which releases are affected. Administrators should review their current Hermes Agent version and consider testing for the route name parsing logic described in the CVE report.

Exploitation requires remote submission of a webhook with a crafted route name and possession of a valid target HMAC. The flaw resides in the gateway logic, so a remote attacker can trigger it from any client that can send a webhook request. The CVSS score of 7.5 indicates high severity, the EPSS score is < 1%, and the CVE is not listed in the CISA KEV catalog, yet the unauthorized elevation of privilege and access to toolsets clearly represents a high‑severity risk.