Critical Claude Code Flaw Silently Bypasses User-Configured Security Rules

Anthropic’s flagship AI coding agent, Claude Code, was recently discovered to contain a critical security flaw that silently bypasses developer-configured safety rules.