Exploring Dynamic Skills in Claude 🚀

In this live stream, Kewin Wereszczynski and Jan Jaworski explore **Dynamic Skills in Claude** through real, unscripted experiments.

The session focuses on how Claude Skills work in practice when paired with AI agents inside a React Native workflow. We test skills against a real codebase, experiment with Expo upgrades, review code using best-practice skills, and look at how context, trust, and supervision affect agent-driven development.

Topics covered include:

- How Claude Skills guide AI agents in real projects
- Context limits, tokens, and breaking work into smaller tasks
- Using skills to review React Native code quality
- Attempting an Expo SDK upgrade with agent assistance
- Creating custom skills and what “human-like” output reveals about AI patterns

This stream is less about polished demos and more about what actually happens when developers try to use AI tools day to day.

See you at other Callstack live streams and webinars ✨ https://clstk.com/4qHxvHr
Follow Callstack on X 🐦 https://x.com/callstackio