A security vulnerability in the AI-powered code editor Cursor can allow malicious code execution when opening compromised repositories. Experts recommend enabling Workspace Trust and auditing repositories to mitigate risks. #Cursor #WorkspaceTrust
Keypoints
- An out-of-the-box security setting in Cursor is disabled by default, exposing users to code execution risks.
- Malicious GitHub repositories can include hidden commands that auto-execute upon folder opening.
- Disabling Workspace Trust increases the risk of leaks, credential theft, and system compromise.
- Threat actors are exploiting prompt injections and jailbreaks in AI coding tools like Claude Code and Windsurf.
- Multiple vulnerabilities including WebSocket bypass, SQL injection, and path traversal pose additional risks to AI development environments.
Read More: https://thehackernews.com/2025/09/cursor-ai-code-editor-flaw-enables.html