/

CVE-2026-75411

Severity:

CRITICAL RISK

(9.8)

Code Injection (CWE-94) in JeecgBoot AI Flow Module

CVE-2026-75411

Deferred

Vulnerability type

Category:

Code Injection

CWE:

CWE-94

Impact:

Remote Code Execution

Severity:

CRITICAL RISK

(9.8)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

JeecgBoot v3.9.2 contains a critical code injection vulnerability in the AI Flow CodeNode component, which supports execution of Groovy scripts. An attacker who can influence or submit CodeNode content may bypass blacklist-based checks and execute arbitrary commands on the server.

Affected context

The vulnerability occurs in JeecgBoot deployments where the AI Flow module exposes the CodeNode component and permits user-controlled or insufficiently trusted Groovy script content to be processed. The risk is increased when the CodeNode is reachable by untrusted users or when the application process has broad operating-system permissions.

Exploit preconditions

  • Attack surface: AI Flow CodeNode script execution functionality
  • Required access: Ability to create, modify, or trigger a flow containing CodeNode content
  • Trigger point: Groovy script evaluation by the CodeNode component
  • Why exploitation is possible: Security checks rely on a blacklist that does not reliably constrain Groovy's dynamic language features, including indirect references and reflective behavior

Why this happens

The underlying flaw is the use of blacklist-based filtering to secure a component that evaluates dynamic Groovy code. Blacklists are incomplete for expressive languages because equivalent operations can be represented indirectly through dynamic resolution, string construction, or reflection. This behavior maps to CWE-94: Improper Control of Generation of Code, because attacker-influenced input can be interpreted and executed as program logic.

Potential impact

  • Attackers may execute arbitrary operating-system commands
  • The application server may be compromised
  • Sensitive application data or credentials could be accessed
  • Attackers could alter workflows, applications, or stored data
  • The compromised host may be used for lateral movement or further abuse

General mitigation guidance

Mitigation generally focuses on eliminating unnecessary server-side script execution and reducing trust in dynamically supplied code.

Common defensive patterns include:

  • Disable or remove CodeNode execution where it is not required
  • Replace unrestricted Groovy evaluation with a constrained declarative workflow model
  • Use an allowlist of narrowly defined operations rather than blacklist filtering
  • Isolate script execution in a dedicated, strongly sandboxed environment
  • Run the application with least-privilege operating-system and service permissions
  • Restrict access to AI Flow authoring and execution functions
  • Apply resource, network, filesystem, and process isolation to execution workers

How teams detect issues like this

Organizations can assess and detect this class of issue through:

  • Static analysis for dynamic Groovy evaluation and blacklist-based security checks
  • Code review of CodeNode handling, script compilation, reflection, and class resolution
  • Runtime monitoring for unexpected child processes, command interpreters, or outbound connections from the application
  • Audit logging of flow creation, modification, publication, and execution events
  • Policy enforcement that prevents untrusted users from submitting executable script content
  • Automated security review tools that flag server-side code evaluation and weak sandboxing patterns

Sources

SourceLink
github.comgithub.com/jeecgboot/JeecgBoot/issues/9691

Frequently asked

questions

It may be remotely exploitable when the AI Flow functionality is exposed to a user who can create, modify, or execute CodeNode workflows. The exact scope depends on deployment permissions and access controls.

Usually, some authenticated workflow-authoring or execution privilege may be required, but deployments that expose these capabilities broadly could increase the risk.

The safest pattern is to avoid evaluating attacker-influenced Groovy code and replace it with constrained, allowlisted operations. If scripting is unavoidable, it should be isolated with strong sandboxing and least privilege.

Treat dynamic scripts as executable code rather than ordinary input, enforce strict authorization around workflow operations, use allowlists instead of blacklists, and monitor execution environments for abnormal process and network activity.

Find out what's already
exploitable in your codebase.