Introducing Claude Code Security, which scans your code for vulnerabilities and suggests fixes

Anthropic has released a research preview of Claude Code Security , a security solution that scans code for vulnerabilities and uses AI to fix them.

Claude Code Security | Anthropic by Claude

Click 'Security Scan' in the upper left corner of the screen.

Click 'Start a scan' to begin scanning the code in your GitHub repository.

A list of vulnerabilities is displayed. Claude Code Security searches for vulnerabilities not by pattern matching, but by 'reading and inferring from the code' like a human security researcher, so it can find vulnerabilities that are difficult to detect with existing automated testing tools. Discovered vulnerabilities also undergo a multi-stage verification process, where false positives are eliminated and a severity level is assigned before being displayed on the dashboard.

Detailed descriptions of each vulnerability are also available.

Click 'Suggest fix' at the bottom right of the screen to generate a suggested fix.

An example of the proposed amendment is as follows:

The fix will not be applied automatically, but only when a human clicks 'Create PR'.

Claude Code Security is available as a limited research preview to Enterprise and Team subscribers, and open source software maintainers are offered free access. You can sign up for the Claude Code Security waitlist at the link below.

Join the waitlist for Claude Code Security
https://claude.com/contact-sales/security