CVE-2026-108667 in JeecgBoot
Summary
by MITRE • 10/11/2026
JeecgBoot through 3.9.5 contains a missing authorization vulnerability that allows low-privileged authenticated users to restore deleted AI prompt templates by calling the revertRecycleBin endpoint. Attackers can send PUT requests to /airag/prompts/revertRecycleBin with chosen template ids to clear deleted flags, undoing administrator removals.
If you want to get best quality of vulnerability data, you may have to visit VulDB.
Analysis
by VulDB Data Team • 10/11/2026
The vulnerability identified in JeecgBoot versions up to 3.9.5 represents a critical failure in access control mechanisms within the AI prompt management module. This flaw is classified as an Insecure Direct Object Reference or Broken Access Control issue, specifically falling under CWE-284 which covers Improper Access Control. The core of the problem lies in the server-side validation logic for the revertRecycleBin endpoint located at /airag/prompts/revertRecycleBin. While the application correctly authenticates users and restricts certain administrative functions to high-privilege accounts, it fails to verify whether the authenticated user has the necessary permissions to execute actions on specific data objects, such as restoring deleted records. This oversight allows any authenticated user with low-level privileges to interact directly with internal API endpoints that are intended for system administrators only.
From a technical perspective, the vulnerability exploits the PUT method associated with the revertRecycleBin endpoint. When an attacker sends a request to this URL and provides specific template identifiers in the payload, the backend processes the command without checking if the requester is authorized to modify or restore those particular items. The application logic assumes that because the user is logged in, they have sufficient rights to perform data restoration operations. This lack of object-level authorization check means that an attacker can manipulate the state of AI prompt templates by clearing their deleted flags effectively undoing administrative decisions to remove sensitive or obsolete configurations. This behavior aligns with MITRE ATT&CK technique T1078 Valid Accounts, where attackers leverage legitimate credentials to gain unauthorized access and perform actions within the system boundaries.
The operational impact of this vulnerability is significant for organizations relying on JeecgBoot for their AI-driven applications. By restoring deleted prompt templates, an attacker can reintroduce malicious or manipulated instructions into the active environment. This could lead to data exfiltration if the prompts are designed to extract sensitive information from users interacting with the AI system. Furthermore it enables potential denial of service by disrupting intended workflows through unauthorized configuration changes. It also undermines audit trails and compliance efforts since actions taken by low-privileged accounts bypass standard administrative oversight mechanisms. The ability to restore deleted items effectively negates data retention policies and security controls designed to limit exposure after a removal event.
To mitigate this vulnerability, developers must implement strict role-based access control checks on all endpoints that modify critical system configurations or data states. Specifically the revertRecycleBin endpoint should verify that the requesting user possesses an administrative role before processing any restoration requests. Additionally implementing comprehensive logging and monitoring for such operations can help detect unauthorized attempts to restore deleted items in real time. Regular security audits focusing on authorization logic across API endpoints are essential to prevent similar flaws from persisting in future releases. Users of JeecgBoot should upgrade to a patched version that addresses this access control deficiency as soon as possible to maintain the integrity and confidentiality of their AI prompt infrastructure.