Executive brief
A security vulnerability has been identified in the CowAgent component of the chatgpt-on-wechat project, which is used to integrate AI chatbots with the WeChat messaging platform. The flaw allows unauthorized individuals to bypass security checks in the Agent Mode Service. This could lead to unauthorized access to the service, potentially compromising chatbot operations or exposing sensitive configuration data.
Technical details
The vulnerability is classified as Improper Authentication (CWE-287) and Missing Authentication for Critical Function (CWE-306) within the Agent Mode Service of CowAgent. The flaw stems from a failure to enforce authentication requirements on certain service functions, allowing a remote attacker to initiate requests without providing valid credentials. This can result in unauthorized manipulation of the agent or access to its underlying functions. The exploit is reportedly public, and as of the advisory date, the maintainers have not yet released a patch or responded to the issue report.
Affected products
- zhayujie chatgpt-on-wechat CowAgent up to 2.0.4
Timeline
- 2026-04-12: disclosed: Vulnerability reported and public exploit noted.
- 2026-04-12: advisory: CVE-2026-6129 published.