Skill
Skills are a technical feature in Claude that enable AI agents to perform specific, well-defined tasks by leveraging structured function definitions. They function as a bridge between an agent’s reasoning capabilities and external systems or computational operations, allowing Claude to understand what actions are available and when to invoke them appropriately.
Implementation and Function Calling
Skills are implemented through function calling, a mechanism that defines the parameters, expected inputs, and outputs of available operations. When a skill is defined, Claude receives a structured schema that describes what the function does, what arguments it accepts, and what results it will return. This allows Claude to reason about whether and when a particular skill should be used to accomplish a given task.
Agent Integration
Within agentic workflows, skills enable Claude to move beyond pure reasoning and take concrete actions in external systems. An agent can evaluate its current goal, consider which skills are available, and decide to invoke the appropriate one. The results of that invocation are then fed back into Claude’s reasoning loop, allowing it to assess progress and determine next steps. This iterative process of reasoning, action, and feedback forms the core of agent functionality.