Anthropic has updated its Claude Security tool to provide continuous repository monitoring. Instead of manual scheduling, the system now triggers scans automatically with every new code commit. Upon initial connection, the tool performs a deep baseline scan, followed by targeted checks for each subsequent change.
The updated system tracks individual vulnerabilities until they are resolved. It automatically marks issues as fixed when it detects the corresponding code changes and can reopen them if the same flaw reappears. Projects support up to eight repositories from platforms like GitHub and GitLab, with data integration available via API and Model Context Protocol.
This feature is currently in public beta for Claude Enterprise customers. Eligible organizations can access up to $25,000 in monthly usage credits for two months, valid until December 31, 2026. Claude Security utilizes the full Claude model family, including Claude Mythos.
Since its public beta launch in April 2026, thousands of organizations have used the tool to identify over one million vulnerabilities. Additionally, Anthropic has introduced an opt-in OSS Scanner for open-source projects, which has already surfaced 493 issues across 25 projects.
Anthropic is also testing a parallel beta plugin for Claude Code, enabling developers to perform local, session-specific scans on individual changes or repositories. These updates aim to integrate security checks directly into existing team workflows.
