Anthropic said thousands of organizations have used Claude Security to identify more than one million vulnerabilities since the product entered public beta in April. The company is now adding continuous repository monitoring and other features to the security tool.
The updated service is available in public beta to Claude Enterprise customers. Anthropic said eligible organizations will receive $25,000 in monthly usage credits for two months, for usage through Dec. 31, 2026. The offer is subject to terms listed in the Help Center.
Now loading...
Unlike a scan initiated manually, the new monitoring system tracks code changes, adjusts scan frequency based on risk and complexity, and keeps findings aligned with the latest state of a repository. Teams can connect repositories from GitHub, including GitHub Enterprise Server, GitLab and other providers, and set the branches and code areas to include.
Projects can include several repositories, allowing the system to identify vulnerabilities that span services. It also checks whether reported issues have been addressed in subsequent code changes, marking them fixed when a correction is deployed and reopening them if they appear again.
Teams can add information about their deployment setup, security policies and other environmental controls to a project knowledge base. Claude Security can also generate a threat model for teams to review and modify, then use that context to prioritize findings and avoid flagging risks already mitigated by existing controls.
Each finding includes a severity assessment based on the project’s threat model and an explanation of how an attacker could reach the vulnerable code. Some reports include a diagram showing the potential attack path.
Anthropic said the service is powered by its full model family, including Claude Mythos, and does not require access to the Cyber Verification Program. A new API allows users to manage projects and repositories, review findings, and resolve or reopen them. Webhooks can alert other systems when new findings are reported.
Claude Security is also available as a connector in Claude.ai and Claude Code, allowing engineers to review findings and begin fixes within the same session. New role-based controls let administrators determine who can view projects, triage findings or create projects.
Enterprise administrators can enable the service through the admin console and control team access. More information is available on the Claude Security product page.
