Anthropic Launches Claude Security Plugin to Scan Code for Vulnerabilities

Anthropic has released the Claude Security plugin in beta for Claude Code, enabling developers to run AI-powered vulnerability scans from the terminal or as full-repo analyses. The plugin maps code architecture, traces data flows across files, detects complex context-dependent vulnerabilities (memory corruption, injections, authentication bypasses, logic errors), verifies findings to reduce false positives, and suggests style-matched patches while integrating with existing workflows; it is positioned as a complement to traditional scanners and is available to Claude Code users.