Skills And Plugins

Even with a well-built Project, you still find yourself re-explaining how to do certain tasks: the shape of your weekly report, the rules of your brand voice, the steps in your review checklist. AI Skills solve that problem, and while the name may sound complex, an AI Skill is really just a folder with a collection of files, oftentimes written in plain language. These files give instructions that the AI can read and follow rather than guessing. The separation into folders breaks up the way you work into small, digestible blocks for the AI.

The complexity of Skills can range from human-readable written preferences to instructions and pre-created, programmed tools, such as executables, code files, and scripts that the AI can make use of as part of the skill. The first kind of skill (text only) is as safe as any prompt given to an AI: it risks prompt injection, but it won't run anything on your device. The second includes runnable code and should be treated like installing software.

Across these sessions you build a real Skill from one of your own recurring procedures and you learn what determines whether a Skill works or fails to live up to its potential. You will also see, live, why choosing skills deserves care: we include one deliberately sneaky demo Skill to show you how it can quietly reshape ordinary output, because an installed Skill influences every matching task without announcing itself.

This workshop runs on a deliberate sources policy: only Skills provided by Anthropic and Skills you build yourself in class. No third-party installs. Our policy is chosen to reflect real-world Enterprise best practices for careful curation, learning to evaluate the impact of a Skill or plugin before ever trusting it.

Participants will learn to:

• Explain what a Skill is and how its description makes it trigger at the right moment
• Build a Skill from one of your own recurring procedures and confirm it triggers on a natural request
• Evaluate a Skill or plugin before trusting it