# Claude gets a security scanner for GitHub repos By Steven Van · 2026-05-01 Claude Security enters public beta for Enterprise customers, pairing Claude Mythos 5 scans with adversarial verification and suggested patches. Anthropic's [Claude]() now has a public beta of Claude Security, a version built for security teams and available to Claude Enterprise customers. It scans a connected GitHub repository using Claude Mythos 5, Anthropic's most cyber-capable model, without giving the account direct access to that model. Findings go through what Anthropic calls an adversarial verification pass, where Claude challenges its own results before surfacing them, aimed at cutting false positives. Each validated finding comes back with a confidence rating and a suggested patch that a team can open in Claude Code on the web and fix using the models available in their account. Every patch requires human review and approval before it's applied. - Focuses on high-severity issues such as memory corruption, injection flaws, authentication bypasses, and complex logic errors that pattern-matching scanners tend to miss, by tracing data flows and reading Git history across files. - Scans can be scoped to a specific directory or run on a schedule for ongoing coverage. - Findings can be pushed via webhooks to Slack or Jira, or exported as CSV or Markdown; dismissal decisions carry forward wherever the team works. A separate **Claude Security Plugin for Claude Code** is also in beta, letting teams scan, validate, and patch directly inside Claude Code while keeping code inside their own environment; it uses whatever models are already available in the account rather than Mythos 5. Admins can turn Claude Security on from the admin console. See the [announcement]() for setup details. [![]()]() ![Claude]() Claude AI assistant by Anthropic — write, analyze, code, and create with one of the most capable large language models available. [View Claude →]() [Original source]() [Read Claude gets a security scanner for GitHub repos on Creators Toolbox]() --- Canonical source: https://creatorstoolbox.com/blog/claude-claude-security-public-beta