Anthropic makes Claude Security continuously scan code repositories
The public beta for Claude Enterprise also rechecks whether previously fixed vulnerabilities stay fixed after code changes.
Anthropic has updated Claude Security so it continuously checks connected code repositories without a person starting each scan, according to digitaltoday.co.kr. The update is available as a public beta for Claude Enterprise customers.
The service now also verifies whether vulnerabilities it previously flagged remain fixed after developers change code. Issues that are resolved are marked as resolved, and revert to unresolved if the problem reappears.
Users set up Claude Security by creating a project, connecting a code repository and specifying the range of code to scan. GitHub connects through the Claude GitHub app, and GitHub Enterprise Server is also supported. Other hosting services such as GitLab connect using an access token.
Anthropic said the security analysis is carried out by its models, including Claude Mythos.
Earlier beta
Anthropic launched the Claude Security public beta in April. Since then, the company says thousands of organizations have found more than 1 million vulnerabilities with the service, a figure that has not been independently verified.
Quick answers
Who can use the Claude Security update?
It is available as a public beta for Claude Enterprise customers.
Which code repositories does Claude Security support?
GitHub connects through the Claude GitHub app, GitHub Enterprise Server is also supported, and other services such as GitLab connect with an access token.
Which models analyze the code?
Anthropic said its models, including Claude Mythos, perform the security analysis.