Malicious .git Configs Can Make Claude, Codex, Cursor, and Other AI Agents Run Attacker Code
By [email protected] (The Hacker News)
AI Summary
Manifold Security disclosed eight security flaws across seven command-line AI coding agents — including Claude, OpenAI Codex, and Cursor — where malicious .git configuration files can execute attacker-supplied commands on a developer's machine at the user's privilege level, outside any sandbox and without an approval prompt. Four of the eight vulnerabilities remained unpatched at time of publication. The attack vector requires only that a developer clone or interact with a malicious repository, making this a realistic supply chain risk for development teams adopting AI coding assistants.
Relevance score: 78.0/100
Expert cybersecurity solutions to safeguard your organization from evolving threats.
Get Protected →