Last released Jul 8, 2026
A read-only security scanner for AI agent skills, rule files, and MCP configurations. Detects prompt injections, hidden instructions, and dangerous permissions before your agent obeys them.
Supported by
