About AgentSkillz Academy

AgentSkillz Academy is a content site about AI agent skills. The site's framing of "skill" is a markdown instruction file that tells an agent what to do, step by step. It's a recipe. A separate concept from a "tool," which is the function the agent calls to actually do something. Most of the confusion in AI development comes from conflating those two things, so this site separates them deliberately.

Two audiences. Developers building agent systems get design patterns, code examples, and a growing skills library of ready-to-use skill files. Everyday AI users get practical guides for getting more out of the tools they already have. The same site, two reading paths.

Where to start

If you're new to all of this, in this order:

  1. Getting started with agent skills — the basics of using AI tools well, no developer experience required
  2. Agent skills glossary — one-line definitions for skill, tool, MCP, plugin, and the rest of the vocabulary
  3. What are agent skills and why they matter — the longer essay version of the glossary's first entry
  4. What's the difference between an AI skill, tool, plugin, and integration? — the deeper treatment of how the layers fit together

What's here

Four categories. Each one organized around a different question.

  • Fundamentals: what skills are, how to design them, when not to use them, why testing matters, and how to think about the security implications
  • Building: design patterns and code examples for developers building agent systems. Error handling, multi-step workflows, observability, orchestration, MCP
  • Using: practical guides for everyday AI use. Writing, research, organization, learning, freelancing, parenting, meeting prep, study, job searching
  • Skills Library: ready-to-use skill files you can copy into your project. PR review, commit messages, changelogs, release notes, incident triage, dependency updates, onboarding checklists, test writing, documentation

The approach

Platform-agnostic where possible. The patterns of good skill design apply whether you're working with Claude, ChatGPT, Gemini, a homegrown agent, or something that didn't exist when this article was written. The site references specific products by name when the details matter, and stays generic where they don't.

Honest about what doesn't work. There's a culture of breathless promotion in AI content that this site explicitly avoids. Articles flag what fails as well as what works. Demos get evaluated as trailers, not movies. When a tool isn't the right answer, the article says don't use the tool.

Practical for non-developers. You don't need to know how to code to get dramatically more value from AI. The using-category articles are written for people who use AI to do their actual jobs, not people who write the AI.

Stay connected

Subscribe to the RSS feed to see new articles and skill files as they ship. Need a reader? Try Feedly, Inoreader, or NetNewsWire.