SkillHub is an enterprise-grade, open-source agent skill registry — a self-hosted platform where teams publish, discover, and manage reusable skill packages across their organization. Think of it as a private npm or PyPI, purpose-built for AI agent skills: you upload a skill package to a namespace, others find it through search or install it via CLI, and the entire lifecycle — from first upload through review, versioning, and governance — is handled within a single, firewall-friendly deployment. Licensed under Apache 2.0, SkillHub ships with a one-command local stack, a CLI-first workflow, and the same polish you'd expect from a public registry, but with full data sovereignty on your own infrastructure.
What SkillHub Solves
In the AI Agent era, every team accumulates skill packages — prompt templates, tool integrations, workflow definitions — but they scatter across local machines, Git repos, and internal wikis. Team members struggle to discover each other's work, let alone reuse proven capabilities. SkillHub provides a private, governed, easy-to-use registry that turns isolated skills into shared organizational assets. The platform enforces review workflows, role-based access, and audit logging so that enterprises can distribute skills at scale without sacrificing compliance or control.
Sources: docs/skillhub/introduction.md
Architecture at a Glance
SkillHub follows a design — business domains are cleanly separated into modules, but everything ships as a single deployable unit. This keeps operational complexity low for the current scale while preserving the option to extract services later. The dependency rule is strict: the domain module sits at the center and depends on nothing; infrastructure implements domain interfaces; the app module wires everything together via Spring DI.