Researchers discovered a critical remote code execution vulnerability affecting major AI coding agents that allows attackers to execute arbitrary code without user interaction through a plugin-based attack vector.
Summary is Secursion's own; full text lives at the source. Attribution preserved.