This curriculum is designed for professionals and assumes you have a solid background in software engineering, generative AI, and experience working with AI coding agents like Claude Code, Cursor, or similar tools. We will be building on top of Claude Code’s extensibility system, but the concepts apply to any agentic coding platform.
This course is for power users who want to go beyond prompting and teach their AI agents new capabilities, giving them domain expertise, custom tools, and reusable workflows they can execute autonomously.
What You Will Learn in Agent Skills
Each skill is a small folder containing everything the agent needs to perform a specific task. A markdown file describes the workflow and decision logic. Executable scripts give the agent hands to act like call APIs, process files, run queries. Configuration files and data sources provide the context it needs to act correctly. When the agent encounters a matching task, it loads the skill automatically, follows the instructions, and delivers expert-level results without any manual prompting from you.
AgentSkills/
- ├── web-search/
- │ ├── SKILL
- │ ├── search.ts
- │ └── sources.json
- ├── code-runner/
- │ ├── SKILL. md
- │ ├── execute.ts
- │ └── sandbox.yaml
- └── README
This course provides a practical dive into , real-world skill building:
- Skill Architecture: Understand the anatomy of a well-structured skill — from markdown instructions to executable code to configuration files and why each piece matters.
- Prompt Engineering for Skills: Master the art of writing SKILL. md files that are clear, unambiguous, and give your agent expert-level behavior in any domain.
