817 Cybersecurity Skills for AI Agents: Framework Mapped

The GitHub repository mukul975/Anthropic-Cybersecurity-Skills publishes 817 structured security skills for AI agents, mapped to MITRE ATT&CK and NIST CSF 2.0, with compatibility for Claude Code and Gemini CLI.

The operational value is the standardization of security workflow logic into loadable artifacts. Previously, agent security behavior was implicit and varied by prompt or tool. This repo converts that tacit knowledge into explicit, versioned, and framework-auditable skills. For builders, integrating these skills reduces the cost of securing agent operations across the development lifecycle by eliminating bespoke security engineering per deployment. It also creates a new audit surface: security posture can now be reviewed as code, compared against baseline frameworks, and updated via repository state instead of manual review. The second-order effect is a shift toward security commoditization for agents—firms will rely on maintained skill libraries rather than internal expertise, making the framework mapping and skill provenance the primary differentiator in agent security infrastructure. This signals that security for AI agents is moving from philosophy to package management.