Agent Skills Hub

Agent Skills Hub is an open-source directory built for developers who need to find, evaluate, and deploy AI agent tools without guessing about quality or safety. Instead of piecing together information scattered across the ecosystem, you get a single place to discover and compare Claude skills, Model Context Protocol (MCP) servers, and other agent tools before they ever reach your production environment.

What sets Agent Skills Hub apart is its automated evaluation pipeline. It continuously crawls the ecosystem, so new entries keep flowing in on their own. Every entry is quality-scored across 10 dimensions, giving you a consistent, side-by-side way to judge which tools are actually worth adopting.

Security is treated as a first-class concern. Each entry runs through security audits designed to guard against common vulnerabilities such as prompt injection, credential exposure, and sandbox leaks—risks that are easy to miss when you are wiring third-party tools into an agent.

The directory is refreshed every 8 hours, so its scores and audits reflect a current view of the ecosystem rather than a one-time snapshot. With its security-graded trust layer, Agent Skills Hub helps teams put AI agents safely into production while keeping up with compliance standards like SOC 2, ISO/IEC 42001, and the EU AI Act.