Anthropic Launches Free Claude Code Terminal Plugin to Detect Security Vulnerabilities

Anthropic has released a free "security-guidance" plugin for Claude Code that continuously reviews AI-generated code in-terminal using a three-layer approach (per-edit pattern scanner, end-of-turn Claude review, and deeper commit-level agentic review). The plugin flags common weaknesses (injection, unsafe deserialization, DOM sinks, workflow risks), integrates with developer workflows without blocking commits, supports repository-specific rules and patterns, and is positioned as an early, extensible line of defense within a broader DevSecOps toolchain.