An agent is only as capable as the tools and skills you give it, and the good news is you can give it new ones through configuration, not code. This lesson is about authoring a skill so your agent can do more.

Think of a skill as a packaged ability your agent can reach for, a defined thing it knows how to do, with a name, a description of when to use it, and the details of what it does. When your agent faces a task, it looks at the skills it has and picks the relevant one, the same way you would reach for the right tool from a toolbox. Authoring a skill is really just describing that ability clearly enough that the agent knows both what it does and when it applies.

The craft is in the description, and it mirrors everything you have learned about briefing. A vague skill description means the agent uses it at the wrong times or misses it when it should apply; a clear, specific one means it fires exactly when intended. You will author a skill by writing that clear description and giving the agent what it needs to perform the ability, then, and this is the part people skip, testing it: give the agent a task that should trigger the skill and confirm it does, then one that should not and confirm it does not. That check is how you know the skill is defined well rather than just defined.

The mindset to carry: extending an agent is not a coding task, it is a describing task, and you are already good at describing because that is the whole course. Every skill you add is another thing your agent can do, authored the same way you brief every other tool, with clarity and a test.