Skip to main content
Vulnerability Database/CVE-2026-108664

CVE-2026-108664: JeecgBoot Authorization Bypass Vulnerability

CVE-2026-108664 is an authorization bypass flaw in JeecgBoot through version 3.9.5 allowing low-privileged users to access AI prompt templates. This article covers technical details, affected versions, and mitigation.

Published:

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

CVE-2026-108664 Overview

CVE-2026-108664 is a missing authorization vulnerability [CWE-862] in JeecgBoot through version 3.9.5. The flaw resides in the AiragPromptsControllerqueryById handler, which fails to verify ownership or role before returning AI prompt records. Any authenticated low-privileged user can read prompt templates belonging to administrators or other tenants.

Attackers enumerate prompt identifiers through the unguarded /airag/prompts/list endpoint, then request each record via queryById. The response discloses prompt text, model parameters, and creator metadata. This exposes proprietary prompt engineering, model configurations, and internal workflows to unauthorized readers.

Critical Impact

Authenticated users with minimal privileges can harvest every AI prompt template stored in JeecgBoot, including administrator-created system prompts and model tuning parameters.

Affected Products

  • JeecgBoot versions up to and including 3.9.5
  • jeecg-boot-module-airag module containing AiragPromptsController
  • Deployments exposing the /airag/prompts/ endpoints to authenticated users

Discovery Timeline

  • 2026-10-10 - CVE-2026-108664 published to NVD
  • 2026-10-13 - Last updated in NVD database

Technical Details for CVE-2026-108664

Vulnerability Analysis

The vulnerability stems from the queryById method of AiragPromptsController in the jeecg-boot-module-airag module. The handler accepts an id parameter and returns the corresponding prompt record without checking whether the authenticated caller owns the record or holds an administrative role. JeecgBoot's framework enforces authentication for the endpoint but defers business-level authorization to each controller, and this handler omits that check.

The companion /airag/prompts/list endpoint compounds the exposure. It returns prompt identifiers without filtering by tenant or creator, giving attackers the ids needed to drive the queryById lookup. The combined behavior turns a single low-privileged account into a reader of every stored prompt.

Disclosed fields include the raw prompt template text, selected model, temperature and token parameters, and creator identity. Organizations that embed business logic, retrieval-augmented generation instructions, or safety rules inside prompts lose that intellectual property to any user with login credentials.

Root Cause

The root cause is a missing authorization check [CWE-862] in the queryById handler. The method retrieves records by primary key without evaluating the caller's identity against the record's create_by field or any role-based access policy. Reference the vulnerable code at the JeecgBoot AiragPromptsController source.

Attack Vector

Exploitation requires network access to the JeecgBoot application and a valid low-privileged account. The attacker authenticates, calls /airag/prompts/list to enumerate identifiers, then iterates /airag/prompts/queryById?id=<id> to retrieve each record. A proof-of-concept script is published in the AnkesKasty cve-request-poc repository.

No verified code examples are reproduced here. See the VulnCheck advisory for the full technical write-up.

Detect Identity Attacks the Moment Authentication Fails

Singularity™ Identity surfaces credential misuse, lateral movement, and privilege escalation in real time, so you can contain the breach before it spreads.

Stop Identity Attacks

Detection Methods for CVE-2026-108664

Indicators of Compromise

  • Sequential or high-volume requests to /airag/prompts/queryById originating from a single authenticated session
  • Requests to /airag/prompts/list followed by rapid queryById lookups across non-contiguous identifiers
  • Access to prompt records whose create_by value does not match the requesting user
  • Low-privileged accounts generating unusual read traffic against the airag module

Detection Strategies

  • Correlate application access logs between /airag/prompts/list and /airag/prompts/queryById to identify enumeration patterns
  • Baseline per-user request rates against the airag endpoints and alert on deviations
  • Compare the authenticated user id against the create_by field returned in prompt responses to flag cross-user reads

Monitoring Recommendations

  • Enable verbose access logging on all /airag/prompts/* routes with user id, source IP, and queried record id
  • Forward JeecgBoot application logs to a SIEM for pattern analysis across sessions
  • Monitor outbound traffic from accounts that access prompt data for signs of exfiltration

How to Mitigate CVE-2026-108664

Immediate Actions Required

  • Restrict access to the /airag/prompts/ endpoints at the reverse proxy or API gateway to administrative roles until a patch is deployed
  • Audit the sys_user table for unused or low-trust accounts and disable any that are not required
  • Review recent application logs for enumeration of the queryById endpoint and rotate any sensitive prompt content that may have been disclosed

Patch Information

No fixed version is listed in the NVD record at time of publication. Monitor the JeecgBoot repository and the VulnCheck advisory for upstream fixes. Operators running 3.9.5 or earlier should plan to upgrade once a patched release is available.

Workarounds

  • Add a server-side authorization filter that compares the authenticated user id to the prompt's create_by field before returning data
  • Restrict /airag/prompts/list and /airag/prompts/queryById to administrator roles using JeecgBoot's @RequiresRoles or Shiro permission annotations
  • Block external access to the airag module paths at the web application firewall for users outside a defined allow list
bash
# Example nginx restriction for the airag prompts endpoints
location ~ ^/airag/prompts/(list|queryById) {
    allow 10.0.0.0/8;      # internal admin network
    deny  all;
    proxy_pass http://jeecgboot_backend;
}

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.