Claude Code Vulnerability Allows Attackers to Run Commands Through Crafted Deeplinks

Claude Code contained a deeplink parsing vulnerability where an eager argv parser misinterpreted a "--settings=..." sequence inside the claude-cli://open?q parameter as a top-level flag, allowing attackers to smuggle a JSON settings object that defines hooks to execute arbitrary commands during session startup; the flaw (demonstrated by a macOS PoC) is fixed in v2.1.118, and defenders should update and treat untrusted claude-cli:// links as dangerous.