GLOSSARY /LESS, BUT BETTER

What Is an AI Skill?

An AI Skill is a reusable set of instructions and resources for handling a particular kind of task.

THE 10-SECOND ANSWER

A playbook for an AI worker

A Skill is a reusable set of instructions and resources that helps an AI agent handle a certain kind of task.

AI Skill

In plain language

Repeating the same instructions every time you ask an AI to do a familiar job is inefficient. A Skill packages that working method so an agent can use it again when the relevant task appears.

In Anthropic's Agent Skills implementation, for example, a Skill can include instructions and metadata as well as scripts, templates and other resources. The exact format and loading behavior can differ between AI platforms, so “Skill” is not one universal package format everywhere. (Claude Platform)

A Skill is therefore not just a prompt. It can be a reusable bundle that helps an work consistently.

An everyday example

An analogy

01New AI workerCan reason
02SkillA reusable playbook
03Completed taskFollows your method

A capable employee can think, but a company playbook explains how this organization expects a recurring job to be done. This is a simplified editorial analogy, not a formal definition shared by every platform.

Where you will encounter it

You may encounter Skills in Claude Code, agent platforms, coding workflows and installable plugin packages. They are especially relevant when a task has repeatable steps, specialist guidance, templates or supporting scripts.

Should you care

  • General chat users: not essential for everyday conversations.
  • Claude Code and agent users: worth understanding.
  • People who repeat the same workflow: particularly useful.
  • Teams with established processes: useful when the method needs to be reusable and consistent.

How it differs

Skill vs prompt: A prompt is usually an instruction for the current interaction. A Skill is organized for reuse and may include instructions, scripts, templates or supporting resources.

Skill vs : A Skill helps an AI understand how to do the work. MCP focuses on which external tools and data the AI can connect to. One does not automatically replace the other.

Skill vs : A Skill can contribute information to an agent's context when it is used, but context is the broader set of information available for the current task.

Read Skill vs MCP for the practical choice between reusable working guidance and an external tool connection.

Related terms

Continue with these published explanations.

Sources & last checked

Sources supplied by the reviewed editorial pack. The check date records our pack cross-check, not a new live verification of these websites. Examples illustrate the concepts.

Official documentation

01
Agent Skills overview — Claude Platform Docsplatform.claude.com

Last checked: September 3, 2026