This is a comprehensive security reference that teaches Claude the OWASP Top 10:2025, plus the LLM and Agentic AI security frameworks. It's built for code reviews and security-sensitive development work. The skill includes quick-reference tables, concrete before-and-after code examples in Python for common vulnerabilities like SQL injection and broken access control, and specific checklists for AI systems (prompt injection, tool misuse, memory poisoning). What stands out is the coverage of emerging risks like goal hijacking in agents and vector store isolation in RAG systems. Use it when you're implementing auth, handling user input, or building anything that touches LLMs or autonomous agents. It's dense but practical.
npx skills add https://github.com/agamm/claude-code-owasp --skill owasp-security