Agent Skill Library: Build Real, Reusable Skills

Agent Skill Library

The first two courses in this path taught you the format and the engineering discipline behind it. This one teaches by building — no new concepts, just eight real, complete skills, each for a tool or domain worth having expertise packaged for, plus a capstone composing three of them together.

Every pack in this course is meant to be genuinely useful, not a toy example. Take what you build here, adjust it to your own team’s specific conventions, and actually install it.

What You’ll Build

• Eight complete skill packs — Python, SQL, GitHub, Kubernetes, Terraform, AWS, LangGraph, and a banking/financial-services enterprise pack
• A composed agent — the capstone combines three of your eight packs into one working, higher-order capability, applying Skill Composition from the companion course
• A published library — tested, versioned, and documented the way the rest of this path taught you to

How This Course Is Organized

Each lesson builds one complete skill pack, end to end — not a fragment or a sketch. Every pack lesson names which design pattern(s) it uses, includes real, working SKILL.md content for that domain, and covers what to test before you’d trust it in your own workflow. The order roughly follows breadth of applicability: language and data tooling first (Python, SQL), then collaboration and infrastructure (GitHub, Kubernetes, Terraform, AWS), then an agent-framework pack (LangGraph), then one deep enterprise domain pack (banking), before the capstone ties several of them together.

Where This Fits in the Path

This is the third course in the Agent Skills Engineering Path, after Agent Skills Mastery (the format) and Production Agent Skills Engineering (the discipline). This course assumes both — every pack lesson references a specific pattern or practice from one of the earlier two rather than re-teaching it, so you can move quickly and spend the time on the actual skill content instead.