Three popular AI agent frameworks had major vulnerabilities, from SQL injection to path traversal, allowing attackers to gain full remote code execution and access sensitive data. Exploits were publicly disclosed, and patches have been released for each framework.









