What Are Claude Skills? A Plain English Guide

Claude skills are reusable sets of instructions that teach Claude exactly how to do a specific task your way, so you can run it again with a single command instead of re-explaining it every time. If you've ever pasted the same long brief into a chat over and over, a skill is the fix. This guide covers what Claude skills are, how they work under the hood, what you can build with them, and how to make your own, with no coding required.

Skills are the feature that turns Claude Cowork from a clever assistant into something that runs your actual workflows. They're also, in our experience training teams on this, the single biggest unlock most people miss.

Watch: the full walkthrough is in our free Claude Cowork course, which covers skills end to end.

What is a Claude skill?

A skill, sometimes called an agent skill, is a packaged set of instructions that tells Claude how to perform one specific task. You define the process once, save it, and from then on Claude follows your exact method whenever you trigger it.

The easiest way to picture it is a filing cabinet. Without skills, giving Claude lots of capabilities means cramming every guideline you have, for emails, reports, social posts, and everything else, into one enormous prompt. All of it sits in front of Claude at once, which causes confusion and mixed-up instructions. With skills, Claude just knows which folders exist in the cabinet. Ask for a branded email and it opens the email folder, reads those instructions, and gets to work, while everything else stays in the drawer.

That shift matters because it lets a non-technical person hand Claude a repeatable job and trust it to do that job the same way every time. The skill holds your template, your tone, your steps, and your rules.

How do Claude skills work?

Under the hood, every skill has three layers, and they load at different moments. This staged loading is what makes skills efficient rather than a drain on Claude's attention.

The first layer is the metadata: the skill's name and a short description. This is always loaded, so Claude always knows the skill exists and when it's relevant. It costs almost nothing, which is why you can have many skills installed without slowing anything down.

The second layer is the instructions: a longer file with the step-by-step workflow. Claude only reads this when the skill is actually triggered, so the detailed guidance enters its working memory at the moment it's needed and not before.

The third layer is the resources: supporting files like templates, fonts, logos, images, or saved scripts. These load only when a task calls for them. A skill might bundle a dozen reference files, but Claude opens just the one it needs for the job in front of it.

This is called progressive disclosure, and it's the reason skills are both powerful and cheap to run. Instead of working out your process from scratch every time, Claude runs a saved script. That means faster, more consistent results and far less wasted usage.

What can you do with Claude skills?

The value of a skill is reuse: you encode a process once and run it forever. Four benefits stand out.

You stop repeating yourself. Define the workflow a single time and run it with one command, instead of writing the same instructions into every new chat.

You get consistent output. Because the method is locked in, Claude executes the same steps the same way each time, rather than improvising and drifting.

You spend fewer tokens. The expensive trial-and-error happens once, while you build the skill. After that it runs as an efficient script, and the instructions only load when relevant.

You give the whole team that capability. A subject-matter expert can encode how they work, then share it, so a process only one person knew how to do becomes something anyone can run.

Claude skills examples

The clearest way to understand skills is to see real ones. These are skills we built and use at 9x.

A video highlights skill takes a workshop recording and its transcript, identifies the best moments, and cuts five social-ready clips with our branding, even switching between the screen-share and speaker views depending on what's on screen. It saves us one to two hours every week.

A branded email skill reads a draft from Notion and builds the finished email directly in our email tool, using our exact template. Before, only a couple of people knew how to build a branded email. Now anyone on the team can.

A tweet-image skill takes a line of text and produces a polished social image at the right dimensions, with the font auto-sizing to fit. Hand it a spreadsheet of fifteen quotes and it returns fifteen finished images.

The pattern across all three: a job that used to need a specialist, or an afternoon, now runs from a single prompt. We walk through building one of these, an email triage agent, step by step in a separate tutorial. For more inspiration, Anthropic publishes a library of open-source skills you can study and adapt.

How to create a Claude skill

You don't need to write code, and the best approach is the opposite of what most people try. Don't open the skill builder and ask it to "create a skill that does X." Do the task with Claude first.

Walk Claude through your process step by step, because you're the expert and you're teaching it your way of working. Check the output and refine it until it's right. Only then do you save it as a skill. The tool that packages it up is Anthropic's Skill Creator, the skill used to build every other skill, and it even tests what it builds before saving.

Once saved, test the skill in a fresh task. From then on, the work runs on demand. For the full build walkthrough, see our guide to creating your own skills, which takes you through it with a real example.

Claude skills vs plugins (and agents)

A few related terms get muddled, so here's the quick version.

A skill is how Claude does one specific task. A plugin is a shareable bundle of several skills plus the connectors and tools those skills need, which makes plugins ideal for onboarding a whole team at once. An agent is the thing that actually plans and takes action across a task; skills are the instructions an agent follows. If you want the full breakdown of that last distinction, our explainer on skills versus agents covers it.

Skills work across both Claude Cowork and Claude Code, so a skill you build for one is available in the other.

A note on overthinking it

People often assume a skill must be a complex, technical thing. It isn't. As one line from a 9x Live workshop puts it: "Skills are just text. Stop overthinking it." Under the hood, a skill is a markdown file describing when to use it and the steps to follow, plus any assets it needs. You don't write that file by hand; Claude writes it from a task you've already done. The work is in doing the task well once, not in the file itself.

Common questions about Claude skills

Do I need to code to make a skill? No. You do the task with Claude in plain language, and it writes the skill file, including any scripts.

How many skills can I have? As many as you like. Only a short description per skill is always loaded, so they don't slow Claude down.

Do skills work in both Cowork and Claude Code? Yes, a skill built in one is available in the other.

Where are skills stored? As folders Claude can read, each with a SKILL.md and any resources. You can download one to share it.

Start building your own skills

Claude skills are the difference between using Claude and operating with it. Start by noticing the tasks you explain to Claude more than once, because each of those is a skill waiting to be built. Do the task together first, save it, and you'll never explain it again.

If you want to learn this properly, with the skills and workflows built around how your team actually works, our hands-on Claude Cowork training takes you from your first task to a working library of skills your whole team can run.