Claude Mythos 5 in Claude Security (Public Beta)
Enterprise customers can point Claude at a GitHub repository and get back a list of security flaws with suggested fixes, using Anthropic's security-specialized model.
Details
- Scans a connected GitHub repository, tracing how data moves across files and how components interact, then returns each finding with a CWE category (the industry's standard label for a type of flaw), a confidence rating, a severity rating, and a suggested fix
- Runs on Claude Mythos 5, Anthropic's security-specialized model, with no separate model access to request
- Billed as standard token usage on the existing Enterprise plan rather than as a paid add-on
- Admins switch it on in the admin console, and people use it at claude.ai/security
- Anthropic also committed $35 million in Claude credits to open-source security organizations through a new Defender Advantage Fund
Best for
security reviews of a codebase you own, where you want flaws found and patched rather than only flagged
When to use it
you are on Claude Enterprise and a repository has never had a proper security pass
Who benefits
security engineers and engineering leads at companies on Claude Enterprise
Join the newsletter to get ahead with AI.
I share strategies, new modules, free guides, and tools to get you better at using AI in your work.
Leading a team? See team programs →