Description

In JetBrains IntelliJ IDEA before 2026.3, 2026.2.2 code execution was possible in Safe Mode via project-controlled JVM options of the build process

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "7.8",
  "pubDate": "2026-11-06T17:17:24.430Z",
  "pubdate": "2026-11-06T17:17:24.430Z",
  "executiveSummary": "A critical security vulnerability exists in JetBrains IntelliJ IDEA that allows for arbitrary code execution while the application is running in Safe Mode.\nThe vulnerability resides in the handling of project-controlled JVM options during the build process, which are improperly validated or restricted.\nBy manipulating specific project configuration files, an attacker can influence the JVM startup parameters used by the build process to inject malicious instructions.\nThis flaw effectively bypasses the security protections intended by Safe Mode, which is designed to limit the functionality of untrusted projects.\nSuccessful exploitation grants an attacker the ability to execute arbitrary code with the privileges of the IntelliJ IDEA process, posing a significant risk to the developer's local environment and potentially sensitive credentials or source code stored therein.\nThe vulnerability affects IntelliJ IDEA versions prior to 2026.3 and 2026.2.2.\nExploitation requires the victim to open a maliciously crafted project within the affected IDE.",
  "technicalDetails": "The vulnerability stems from an insecure implementation of the build process configuration mechanism within IntelliJ IDEA. Safe Mode is intended to act as a sandboxing or restriction layer when opening projects from untrusted sources, limiting the ability of the IDE to execute arbitrary code defined within project files.\nHowever, the build system's handling of JVM options—specifically those defined via project-level configurations—fails to implement sufficient sanitization or validation during the build initialization phase. When a project is loaded, the IDE reads configuration files that define how the build process should be instantiated, including custom JVM arguments.\nIf these arguments are under the control of an attacker (e.g., via a shared project configuration file or a repository pull request), the attacker can inject malicious JVM flags. For instance, flags such as -javaagent or other diagnostic options that facilitate arbitrary code execution can be passed to the build process JVM.\nBecause the build process inherits the security context of the parent IntelliJ IDEA process, the injected code executes with the full privileges of the logged-in user. This bypasses the intended Safe Mode restrictions because the build-time environment is not fully isolated from these configuration-driven parameters.\nThe attack flow follows these steps: 1. The attacker creates a project repository containing a modified configuration file that embeds malicious JVM parameters in the build settings. 2. A victim opens this project in a vulnerable version of IntelliJ IDEA. 3. Even if the IDE prompts for Safe Mode, the internal build system processes the malicious configuration when preparing the project build environment. 4. The build process JVM is launched with the attacker-supplied arguments, leading to immediate code execution upon build initialization or triggers. 5. The payload executes within the context of the user, potentially allowing for persistence, data exfiltration, or further lateral movement within the local machine's environment."
}