Anthropic Launches Claude Code Security for Vulnerability Scanning

Hey everyone, big news from Anthropic! They've just unveiled Claude Code Security, a powerful new capability built directly into Claude Code on the web. This tool is set to change how development and security teams tackle code vulnerabilities, moving beyond traditional methods to find those tricky, hard-to-spot issues. You can dive into the full details on the Claude Code Security Announcement.

What it does

Claude Code Security is designed to act like an expert human security researcher, but with the speed and scale of AI. Instead of just scanning for known patterns, it reads and reasons about your codebase, understanding how different components interact and how data flows through your application. This allows it to identify complex, context-dependent vulnerabilities that often slip past rule-based tools.

Here’s the breakdown of its core functionality:

Scans for vulnerabilities: It thoroughly examines your code for potential security flaws.
Suggests targeted patches: Once a vulnerability is found, Claude Code Security doesn't just flag it; it suggests precise software patches for human review.
Intelligent verification: Every finding goes through a multi-stage verification process, where Claude re-examines results to filter false positives. It also assigns severity ratings and provides a confidence score for each finding, helping teams prioritize.

Crucially, no patches are applied without human approval. Claude Code Security identifies problems and offers solutions, but the final decision always rests with the developers, ensuring critical human oversight.

Why it matters

The challenge for security teams has grown immensely: too many vulnerabilities, not enough skilled personnel to address them. Traditional static analysis tools are helpful but often miss sophisticated flaws that require a deep understanding of code logic. This is where Claude Code Security shines.

Leveraging the advanced reasoning capabilities of Claude Opus 4.6, Anthropic's team has already made significant discoveries, finding over 500 previously undetected vulnerabilities in production open-source codebases. These were bugs that had gone unnoticed for years, even with expert review! This level of performance highlights the tool's potential to significantly raise the security baseline across the industry. By putting this powerful AI in the hands of defenders, Anthropic aims to protect code against a new category of AI-enabled attacks.

How to try it

Anthropic is currently offering Claude Code Security in a limited research preview. It's available to Enterprise and Team customers, with expedited access also offered to maintainers of open-source repositories. This collaborative approach allows Anthropic to refine the tool's capabilities responsibly. To learn more about its features and benefits, you can visit the Claude Code Security Product Page or check out Anthropic's homepage for more on their AI advancements.

Try it: https://www.anthropic.com/news/claude-code-security to get started securing your code.