AI Agents: Tools, MCP & Skills
Understand how AI agents take action, connect them to your tools, and keep them under control
AI agents don't just answer questions: they search the web, read your files, run code and take actions on your behalf. This course explains, without requiring any coding, how that actually works. You'll learn the agent loop, tools and function calling, server tools like web search and code execution, the Model Context Protocol (MCP), and skills, plus the permission and safety practices that keep agents from doing damage. Along the way you'll install a real MCP server, write your own skill, and learn to spot risky actions before you approve them.
6 lessons · ~22 min video
First lesson free · then membership
Course content
Module 1
How agents work
The agent loop, how models use tools through function calling, and the server tools (web search, web fetch, code execution) that providers run for you.
Module 2
Extending and controlling agents
Connect agents to your systems with MCP, teach them repeatable processes with skills, and keep them safe with permissions, sandboxing and careful review.