Gate Every PR With Claude Code Security Reviews

AI coding assistants now write a large share of the code that reaches production, and human review has quietly become the bottleneck. Anthropic's answer is a security reviewer that reads your diffs the way an application-security engineer would. The open-source repo behind it, anthropics/claude-code-security-review, has crossed 5,600 GitHub stars, and the two features it ships are becoming the default way teams gate a merge.

What sets this apart from grep-style SAST is context. Claude follows tainted input across functions, understands what the change is trying to do, and filters the noise that makes traditional scanners exhausting to live with. Anthropic reports that its own GitHub Action caught a DNS-rebinding remote code execution bug and an SSRF in internal tools before either pull request was merged.