Post by Arcade.dev

6,783 followers

We built ToolBench to benchmark MCP servers. Today we are launching SkillBench to do the same for agent skills. Skills are how expertise moves between agents. Write one good skill describing a business workflow and every employee that selects it, and every agent that installs it, inherits the workflow. So it’s no surprise that the public ecosystem has exploded to tens of thousands of published skills in just the last few months, but that growth has outpaced any way to judge quality before you adopt. SkillBench has scored 39,000 skills and counting including popular skills from Anthropic, Microsoft, Vercel, and more, each scored across six dimensions that matter most to enterprise businesses including Safety & Malicious Intent, Tool Boundary, and Workflow Quality. Teams deserve transparent scoring before they adopt new skills. Learn about our methodology, check a skill’s score, and submit your own at https://lnkd.in/eCuACSkQ

Post content