Anthropic's new Claude Security tool scans your codebase for flaws - and helps you decide what to fix first

Anthropic has launched Claude Security, an AI-powered vulnerability scanner now in public beta for Enterprise Claude users. Built on the Opus 4.7 model, it scans full repositories, traces data flows across files, and ranks findings by confidence and severity so developers can prioritize critical fixes first. It also connects directly to Claude Code for in-context patching, with integrations from CrowdStrike, Wiz, and Palo Alto Networks.