"AI skills" now means more than one thing.

"AI skills" is a phrase in heavy use, and it doesn't always mean the same thing. Depending on who is speaking, it can refer to human abilities, to a technical format for agents, or to a product feature a vendor calls "skills." This guide separates the three meanings so you can tell them apart.

Human AI skills

The first and most common meaning is human. "AI skills" often describes the abilities people develop to work effectively with AI — writing clear prompts, breaking problems into steps, evaluating outputs, and knowing when to trust a model and when not to.

These are durable, transferable skills. They are about fluency with AI as a tool, and they apply regardless of which product or model you use.

Agent Skills

The second meaning is technical. An Agent Skill is a reusable package of instructions and resources — built around a SKILL.md file — that an AI agent loads when a task matches. This is an open format now supported across multiple agent ecosystems. It is the meaning most relevant to builders and developers.

If that's what you're after, start with What are AI Agent Skills? and the SKILL.md guide.

Product capabilities

The third meaning is commercial. Some vendors use "skills" as a marketing label for features or capabilities inside their products — sometimes overlapping with the Agent Skills format, sometimes not. Here, "skills" is shorthand for "things the product can do," and the specifics vary by vendor.

Why context matters

These meanings overlap just enough to cause confusion. A recruiter posting about "AI skills" almost certainly means human skills. A developer asking about "skills" in a coding agent probably means Agent Skills. A product page advertising "skills" may mean either — or neither.

The practical move is to ask which layer is being discussed:

• People learning to use AI → human skills.
• Files that package reusable agent workflows → Agent Skills.
• Features branded "skills" by a product → vendor capabilities.

How this site uses the term

On AISkills.guru, "skills" and "Agent Skills" refer to the technical format unless we explicitly say otherwise. The field guide covers that category end to end — including Claude and Codex specifics, how skills compare to MCP, and how to build one.

Frequently asked questions

Which meaning is most common? In general usage, human skills. In developer and builder contexts, Agent Skills.

Are Agent Skills the same as the "skills" in my AI product? Sometimes. They use the same open format only where the product explicitly supports it. Check the product's docs.

Do I need Agent Skills to have "AI skills"? No. Human AI skills need no software at all. The two meanings are independent.

Where did the Agent Skills format come from? It is an open format; see agentskills.io for the specification and ecosystem.