In the past year, we've seen rapid advancement of model intelligence and convergence on agent scaffolding. But there's still a gap: agents often lack the domain expertise and specialized knowledge needed for real-world work. We think Skills are the solution—a minimal form factor for packaging procedural knowledge that agents can dynamically load. It's a portable, composable approach to giving one agent capabilities across domains. In this talk, we'll share how we built Skills at Anthropic, the network effects we're observing, and where we believe this leads: agents writing their own Skills from experience. Our thesis: equipping agents for real-world work means building reusable expertise.

Barry: https://twitter.com/barry_zyj
Mahesh: https://twitter.com/MaheshMurag

AI Engineer

Anthropic introduces agent skills, a new paradigm for extending AI agents with domain expertise. Skills are organized folders containing scripts, instructions, and procedural knowledge that agents can progressively load at runtime. This approach treats code as a universal interface, allowing general-purpose agents to gain specialized capabilities without rebuilding separate agents for each domain. Skills complement MCP servers by providing expertise while MCP provides connectivity. The architecture enables non-technical users to create skills, supports versioning and composability, and aims toward continuous learning where agents improve over time by creating and refining their own skills.

Don't Build Agents, Build Skills Instead – Barry Zhang & Mahesh Murag, Anthropic