Cursor Vulnerability Exposes Millions of Developers to Code Execution Threat
A potentially catastrophic vulnerability has been discovered in Cursor, a popular AI-assisted development environment used by over 7 million active users on Windows. The security defect allows attackers to execute malicious code simply by planting a malicious git.exe binary in a repository’s root directory, which can then be automatically executed by Cursor without warning or approval.
This exploit takes advantage of Cursor’s path resolution logic when loading projects, which includes searching for Git binaries in multiple locations, including the workspace itself. If an attacker places a malicious git.exe file at the repository’s root, Cursor will execute it automatically as part of its normal operation, making it possible to run arbitrary code on the victim’s system.
The vulnerability has been disclosed publicly by security researchers Mindgard after they reported it to Cursor in December 2025 and received no response for seven months. The issue was subsequently confirmed through Cursor’s bug bounty program on HackerOne, but again, there has been no indication of a patch or notification to affected users.
This situation highlights the importance of coordinated disclosure, which relies on collaboration between security researchers, vendors, and end-users to ensure that vulnerabilities are addressed promptly. However, in this case, the silence from Cursor is concerning, as it leaves millions of developers vulnerable to attacks.
The vulnerability is straightforward to exploit and does not require any complex or sophisticated techniques. It’s essential for Cursor users to be aware of this risk and take necessary precautions to protect themselves. In the meantime, Cursor should prioritize patching this issue and providing clear guidance to its users on how to mitigate the threat.
As a developer, it’s crucial to be cautious when opening projects from unknown sources and to verify the integrity of your system regularly. This vulnerability serves as a reminder that even popular development tools can have security flaws, emphasizing the need for continuous monitoring and patching.
Source: SecurityWeek — 2026-07-15