Learn to create custom AI agent skills and build working agentic prototypes using assisted coding tools like Cursor or Claude Code. You'll start with an introduction to how skills work: their structure, purpose, and what they can unlock, before sketching out your own MVP idea tied to a project or area of work you care about.

From there, you'll design, build, and run at least one skill locally, seeing firsthand how instructions and scripts shape agent behavior. But building is only half the picture: you'll also collectively stress-test your skills, identifying where they could fail, be manipulated, or cause unintended harm and iterate to make them more robust. Drawing on real-world examples from agentic writing support at QuillBot, the workshop balances practical making with critical thinking about what AI should do, not just what it can do.

You'll leave with a functional prototype running on your machine and a clear framework for designing agentic flows independently going forward.