About
Safe Pkgs provides a robust security layer for AI agents by performing comprehensive package safety checks prior to installation. It acts as a Rust-based MCP server and CLI, offering machine-readable decisions including allow/deny status, a detailed risk score (low, medium, high, critical), human-readable reasons for findings, and essential package metadata. Currently supporting npm and cargo registries, it helps prevent the introduction of vulnerable or malicious dependencies into AI agent environments, backed by audit logs and an SQLite cache for efficient operation.