On September 2, 2026, security researchers at Manifold Security disclosed GitSpawn, a class of vulnerabilities affecting AI coding agents including Claude Code, Codex, Cursor, and Grok. Attackers can exploit malicious repository configurations to execute arbitrary code on developers' machines with user privileges.
No BS check run yet — press ⚖ to extract this story's claims and verify them against independent sources.
Security researchers at Manifold Security disclosed GitSpawn, a class of vulnerabilities that can let a malicious repository make AI coding agents execute attacker-controlled code on a developer’s machine as soon as the project is opened, without a prompt or approval click [heise.de#1][cybersecuritynews.com#1][thehackernews.com#1]. The affected tools named in the reports include Claude Code, Codex
From yesterday: Manifold Security: GitSpawn: A Single Flaw Lets Untrusted Repos Run Code in Claude Code, Codex, Curser and Grok https://www. manifold.security/blog/ai-codi ng-agents-git-hijack More: The Hacker News: Malicious .git Configs Can Make Claude, Codex, Cursor, and Other AI Agents Run Attacker Code https:// thehackernews.com/2026/09/mali cious-git-configs-can-make-claude.html @ thehackern
GitSpawn: A Single Flaw Lets Untrusted Repos Run Code in Claude Code, Codex, Cursor, and Grok - Manifold Security # ClaudeCode # Cursor # Grok # CVE_2026_19592 # CVE_2026_72718 https://www. manifold.security/blog/ai-codi ng-agents-git-hijack
🤖 Manifold Security discloses 8 flaws in 7 AI coding agents (Claude, Codex, Cursor, ...): a malicious .git config in a repository makes the agent run attacker commands on the dev machine — as the user, outside the sandbox, without an approval prompt. 4 of 8 still unpatched. 🔗 https:// thehackernews.com/2026/09/mali cious-git-configs-can-make-claude.html # Exploit # AISecurity # CyberSec
Malicious .git Configs Can Make Claude, Codex, Cursor, and Other AI Agents Run Attacker Code https:// thehackernews.com/2026/09/mali cious-git-configs-can-make-claude.html