/

CVE-2026-67926

Severity:

CRITICAL RISK

(9.8)

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

CVE-2026-67926

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

A critical code injection vulnerability in the JeecgBoot AI Chat Module allows a remote attacker to execute arbitrary code through processing of the files parameter.

The issue becomes risky when attacker-controlled file-related input reaches an execution-capable code path without adequate validation, isolation, or authorization.

Affected context

The vulnerability affects JeecgBoot v3.9.2 deployments that expose the AI Chat Module and process the files parameter through a vulnerable code path. Risk depends on whether the module is remotely reachable, whether the parameter is accepted from untrusted users, and what permissions are available to the application process.

Exploit preconditions

  • Attack surface: Network-accessible JeecgBoot AI Chat Module endpoint
  • Required access: The ability to submit requests containing the files parameter; authentication requirements depend on deployment configuration
  • Trigger point: Processing of the files parameter by the AI Chat Module
  • Why exploitation is possible: Attacker-controlled input may reach an execution-capable operation without sufficient validation, neutralization, or isolation

Why this happens

The underlying bug is an unsafe code-handling pattern in which externally supplied input is permitted to influence executable behavior. This maps to CWE-94: Improper Control of Generation of Code ('Code Injection'), where application data is not adequately separated from code or commands and is subsequently interpreted by the runtime.

Potential impact

  • An attacker could execute arbitrary code in the JeecgBoot application context
  • The application process may be fully compromised
  • Sensitive data and credentials may be exposed
  • The attacker could alter application behavior or stored data
  • Compromise could enable lateral movement, depending on network access and service privileges

General mitigation guidance

Mitigation generally focuses on ensuring that file-related parameters remain data rather than executable instructions.

Common defensive patterns include:

  • Replacing dynamic evaluation or execution with fixed, allowlisted operations
  • Applying strict type, format, size, and content validation to file-related inputs
  • Separating file parsing from command or code execution paths
  • Restricting accepted file types and storage locations
  • Running the application with least-privilege permissions and limiting outbound network access
  • Using authentication and authorization controls appropriate to the module's exposure
  • Isolating file-processing components from sensitive application and host resources

How teams detect issues like this

Organizations can assess for this class of issue through:

  • Static analysis rules targeting dynamic evaluation, reflection, template execution, and command construction from request parameters
  • Code review of all flows from the files parameter into interpreters, scripting engines, process launchers, or generated code
  • Runtime monitoring for unexpected child processes, command interpreters, or unusual file-processing behavior
  • Web application policy enforcement that restricts parameter formats, file types, and request destinations
  • Automated security testing that uses benign markers to verify whether untrusted input reaches execution-capable sinks

Sources

SourceLink
github.comgithub.com/jeecgboot/JeecgBoot/issues/9774

Frequently asked

questions

Yes. The description identifies a remote attack path through the AI Chat Module, although practical exposure depends on endpoint reachability and deployment controls.

Not necessarily. The required access depends on the module's authentication and authorization configuration; any account or network position permitted to submit the affected request may be relevant.

Ensure that the files parameter is treated strictly as validated data and cannot influence dynamic code generation, evaluation, or command execution.

Use allowlisted input schemas, eliminate dynamic execution where possible, apply least privilege, isolate file-processing components, and include code-injection checks in review and testing workflows.

Find out what's already
exploitable in your codebase.