Claude Skills open up a practical way to extend an AI agent’s behavior in real time. By adding context, scripts, or instructions on the fly, you can teach an agent new skills. In this live stream, we’ll explore how this model works in practice and where it fits into real-world systems.

Kewin Wereszczyński and Jan Jaworski will walk through concrete use cases and discuss how Claude Skills can be applied to adapt agents to changing requirements. We’ll look at how dynamic instructions affect behavior, what kinds of problems this approach solves well, and where its limits start to show.

The session will be hands-on. We’ll test capabilities live, experiment with different skill setups, and play around with teaching the agent new behaviors as we go. Expect practical insights, not theory, and a clear picture of how Claude Skills behave under real conditions.

Join us on Thursday, January 29 at 4 PM to see how dynamic agent skills work in practice and what they enable when you start experimenting beyond static prompts.