Hey! Ankur here, and this is the 13th edition of Lazy AI — 5 mins of AI reading to help you stay ahead of the AI curve.

Yesterday, I spoke about what Claude Skills are and why they’re one of the most underrated features in Claude.

Today, I’m getting into the actual setup — and I’ll show you the manual way AND the lazy way to set it up.

(Spoiler: the lazy way is better) 😉

Claude Skills are instruction files you create once and upload to Claude. Whenever you work on something relevant, Claude automatically detects the right Skill and loads it — bringing your preferred context, tone, and workflow into every conversation, without you having to explain yourself again.

If you want to know details, I’m linking yesterday’s newsletter below

Now, how do you actually set one up?

Before anything else, go to Settings > Capabilities and make sure Code Execution and File Creation is turned on. Skills won’t work without this.

Then go to Customize > Skills. This is where you’ll manage everything — Anthropic’s pre-built skills, partner skills, and your own custom ones.

A Skill is essentially a folder containing one core file: a SKILL.md file. It starts with some basic metadata — a name and a description — followed by instructions for Claude.

Here’s where it gets interesting. You can create this two ways.

Go to Customize > Skills and click on “Create a new skill”. Click the “+” icon and write your instructions, or upload a file. The file looks like this:

---name: my-skill-namedescription: A clear description of what this skill does and when to use it---

# Instructions[Everything Claude should know and do when this skill is active]

The description is the most important part — it’s what Claude reads to decide whether to load your Skill for a given task. Be specific. Here’s what that looks like in practice:

A content writing skill might say: “Use this when writing blog posts, newsletters, or LinkedIn content. Tone is conversational and direct. No jargon. First-person. Short sentences. Avoid filler phrases like ‘in today’s world’ or ‘it’s worth noting.’”

A report publishing skill might say: “Use this when creating business reports or client-facing documents. Format as a structured Word document with an executive summary, clear section headers, and a recommendations table at the end. Professional tone. Data before opinion.”

These are just small examples of the description. In practice, the more detailed it is, the better is your output.

But it’s kinda tedious to do it this way. Which is why I prefer option 2.

You don’t need to be an engineer to talk AI. Subscribe to this newsletter — every other day, you’ll understand AI well enough to be the smartest person in the room at work.

Each issue is just 5 minutes — less than the time you spend doomscrolling before bed. Except, this actually moves your career forward. Join 8,000+ subscribers now.

Here’s the thing — you don’t have to write the SKILL.md file yourself. Claude can do it for you.

Open a fresh conversation and say something like:

Claude will interview you. It’ll ask about your role, your audience, your tone, examples of good and bad outputs. Ten minutes later, it hands you a ready-to-use SKILL.md file. You just click on “Copy to your skills” and it’s done. Your skill is ready.

No prompt engineering required here. You just answer questions honestly.

Once uploaded, start a new conversation and do something the Skill is meant to help with. Claude should automatically detect and load it.

If it doesn’t feel right, go back into the file — tighten the description, add more specific instructions, include an example of what “good” looks like. Skills aren’t a one-time setup. Think of them as a living document you improve over time.

The manual way gives you more control. The lazy way gets you something useful in 20 minutes. Either way, once it’s set up, you stop spending the 30 minutes of every conversation re-explaining what you need.

Slowly, you refine your skill, and then few iterations later, it works with just one single prompt. It’s like having an assistant who will do whatever you ask it to, with precision.

That’s all for today, folks! Please implement Claude Skills in your everyday AI work. Trust me, it’s a game-changer!

If this got you thinking, feel free to subscribe. And please share it with someone who uses AI every day but still types the same context into every single conversation.

It takes a lot of time to research and write this newsletter. Sharing it would mean a lot 🙂

See you next time..

Cheers,

Ankur