A missing authorization vulnerability in vmDeshpande AI Agent Automation allows authenticated attackers to access or destroy sensitive cross-tenant data, posing a high risk to data confidentiality and integrity.

The application fails to perform proper authorization checks in the memory controller. While requests are authenticated, the system does not verify that the requested agent or memory identifiers belong to the authenticated user, allowing any authenticated attacker to manipulate data belonging to other tenants.

This vulnerability carries a CVSS score of 8.8, reflecting its high impact on confidentiality, integrity, and availability. Successful exploitation allows unauthorized access to conversation history, agent context, and sensitive task data, which can lead to significant data breaches, regulatory non-compliance, and loss of user trust.

Immediate Action: Upgrade the AI Agent Automation platform to version 0.9.1 or later to implement the necessary authorization logic.

Proactive Monitoring: Review application access logs for unusual patterns of memory access, particularly requests involving identifiers not associated with the active user session.

Compensating Controls: Implement strict API gateway controls or network-level access restrictions to limit the exposure of the memory controller endpoints until the patch can be applied.

Public Exploit Available: Yes, a public proof-of-concept exists on GitHub.

Given the high severity of this vulnerability and the existence of a public proof-of-concept, administrators should prioritize updating to version 0.9.1 immediately. Failure to patch allows for trivial cross-tenant data theft, which could have severe legal and operational consequences for organizations utilizing this automation platform.