Since you're looking at openclaw/clawhub, you're exploring the backbone of how modern autonomous agents gain new capabilities (or "skills"). Here is a breakdown from an engineering perspective