What Are Agent Skills?

Teach Claude Once

Every time you explain your team's coding standards to Claude, you're repeating yourself. Every PR review, you re-describe how you want feedback structured. Every commit message, you remind Claude of your preferred format. Agent Skills fix this: a skill teaches Claude how to do something once, and Claude applies that knowledge automatically whenever it's relevant.

ℹ️

The one-sentence definition

A skill is a folder containing a SKILL.md file with instructions — plus optional scripts and resources — that Claude loads when a task matches. Anthropic describes skills as 'custom onboarding materials that let you package expertise, making Claude a specialist on what matters most to you.'

The rule of thumb is simple: if you find yourself explaining the same thing to Claude repeatedly, that's a skill waiting to be written. Instead of pasting the same checklist into chat over and over, you write it once and Claude reaches for it on its own.

An Open Standard: Write Once, Use Everywhere

Skills aren't a Claude-Code-only trick. They follow the Agent Skills open standard (agentskills.io), a consistent format that works across multiple AI tools. Anthropic's design goal is portability — 'write once, use everywhere': you build a skill a single time and deploy it across the Claude apps, Claude Code, and the API. Claude Code then extends the standard with extra features like invocation control and subagent execution.

One skill, written to the open standard, runs across the Claude apps, Claude Code, and the API — write once, use everywhere.

What a Skill Looks Like

Every skill lives in a SKILL.md file with two parts: YAML frontmatter between '---' markers (metadata that tells Claude when to use the skill) and Markdown content below (the actual instructions Claude follows). The description is how Claude decides whether the skill is relevant to your request.

markdownA minimal skill: frontmatter on top (when to use it), instructions below (what to do).

---
name: pr-review
description: Reviews pull requests for code quality. Use when reviewing
  PRs or checking code changes.
---

When reviewing a pull request:
1. Run `git diff main...HEAD` to see the changes
2. Check for missing error handling and exposed secrets
3. Group feedback by severity: critical, warning, suggestion

⭐

Skills bundle more than text

Beyond instructions, a skill folder can include executable scripts and resources (templates, examples, reference docs). Claude pulls in only what each task needs — so a skill can carry deep expertise without bloating every conversation. We'll build multi-file skills later in the course.

What Skills Are Great For

Skills shine for specialized knowledge that applies to specific tasks — the procedures and standards you'd otherwise re-explain every time.

•Code review standards your team follows
•Commit message and PR description formats you prefer
•Brand guidelines (fonts, colors) for your organization
•Documentation templates for specific kinds of docs
•Debugging checklists for a particular framework

Without a skill	With a skill
Re-paste the same checklist every PR	Write it once; Claude applies it automatically
Instructions clutter every conversation	Loads only when the task matches
Inconsistent results across the team	Shared, consistent standard for everyone

Skills turn repeated instructions into reusable, automatically-applied expertise.

✨

Skills stay efficient because Claude doesn't load them all the time — it uses progressive disclosure. That mechanism is the heart of why skills work, and it's next.

Key Takeaways

✓A skill is a folder with a SKILL.md (instructions) + optional scripts/resources that Claude loads automatically when a task matches.
✓Skills let you teach Claude once instead of repeating the same instructions — 'if you keep explaining the same thing, that's a skill waiting to be written.'
✓Skills follow the Agent Skills open standard (agentskills.io): 'write once, use everywhere' across the Claude apps, Claude Code, and the API.
✓SKILL.md has YAML frontmatter (when to use it — the description drives matching) and Markdown content (the instructions Claude follows).
✓Skills can bundle scripts and resources, not just text, so they carry deep expertise without bloating every conversation.
✓Best uses: code-review standards, commit/PR formats, brand guidelines, doc templates, debugging checklists — anything you'd otherwise re-explain.

Check Your Understanding

Test what you learned in this lesson.

Q1.What is an Agent Skill?

Q2.What does the Agent Skills open standard (agentskills.io) enable?

Q3.Within a SKILL.md file, what is the role of the frontmatter description?

Q4.Which is the best signal that something should become a skill?

Practice This Lesson

Progressive Disclosure: How Claude Loads Skills