Artificial intelligence is accelerating its evolution from "pure large models" to "intelligent agents that can handle tasks." According to AIbase, Anthropic has made a major upgrade to the "Skills" feature in its Claude model this Thursday, aiming to enable AI to assist users more independently in handling specific tasks through structured task training.

The core action of this update is the release of the "Agent Skills" specification as an open standard. This means developers and enterprise users can more easily create, deploy, and share new skills for AI agents. For example, users can teach Claude how to fill out complex forms or browse specific websites simply by adding files that describe the process in a specific folder, enabling AI to perform actions directly rather than just engaging in conversations.

To further expand into the enterprise market, Anthropic also introduced organization-level management features, allowing administrators to centrally manage the skill libraries within their companies. At the same time, through collaborations with well-known office platforms such as Canva, Notion, and Figma, users can directly access these pre-set professional skills. AIbase believes that this move marks a shift in the AI market from a "model competition" to a "use case competition," with AI becoming independent digital employees capable of solving problems on their own.