Definition
A discrete capability extension that gives an AI agent permission to perform a specific action — such as browsing the web, reading files, running code, or calling an external service. Skills are defined in configuration files and are the mechanism through which agents gain real-world reach beyond generating text.
Why it matters
Every skill installed on an AI agent expands what an attacker can cause the agent to do if they manipulate it; auditing, signing, and restricting agent skills is one of the highest-leverage security controls available to enterprises deploying AI agents.