Claude Skills Tutorial: Build, Meta, Pitfalls
Key Points
- The video provides a step‑by‑step tutorial for creating Claude Skills, including how to avoid common mistakes and how to build “meta‑skills” that can be reused to construct other skills.
- Skills act as plugins or extensions that give Claude specialized instructions while reducing prompt length; they can be loaded from local folders, uploaded as zip or the newer *.skill* files, or managed via the API (which requires code execution/file‑creation to be enabled).
- A major source of confusion is that each platform expects a different packaging format: Claude Code looks for folders under */skills*, the web/desktop apps accept zip or *.skill* files, and uploaded zip files must be extracted into the correct local directory to be recognized.
- Despite these quirks, the zip‑based skill format is interoperable—e.g., ChatGPT can open a *.skill* file as a zip—so you can leverage the same skill assets across different AI tools, provided you follow the proper extraction and placement steps.
Sections
- Creating and Sharing Claude Skills - A comprehensive tutorial that walks viewers through building Claude skills, avoiding common mistakes, leveraging meta‑skills, and handling the various file formats and deployment methods.
- Common Claude Skill Pitfalls - The speaker outlines frequent errors such as forgetting to enable code execution, providing vague skill descriptions that misguide Claude, using unclear markdown headers, and suffering from inconsistent versioning, urging developers to be explicit and maintain a single source of truth.
- Building a PowerPoint Chunking Skill - The speaker outlines how to create a skill that automatically checks and splits large PowerPoint requests to avoid exceeding Claude's context window, ensuring smoother generation.
- Automated Skill Generation Walkthrough - The speaker walks through an 11-step automated process that creates a PowerPoint chunking skill, outputs documentation and strategy details, and packages everything into a functional but visually garbled zip file.
- Simplifying Skill Management for Non‑Technical Users - The speaker describes how meta‑skills and built‑in tools such as profilers, gap analyzers, and token‑budget advisors streamline the creation, tracking, and optimization of AI skills, making complex work accessible to non‑technical people across platforms.
- Building Prompt Skills for Testing & Security - The speaker describes creating dedicated testing and security prompt skills, advises naming them after functions rather than teams, and previews a Claude continuous‑learning skill.
Full Transcript
# Claude Skills Tutorial: Build, Meta, Pitfalls **Source:** [https://www.youtube.com/watch?v=WKFFFumnzYI](https://www.youtube.com/watch?v=WKFFFumnzYI) **Duration:** 00:21:40 ## Summary - The video provides a step‑by‑step tutorial for creating Claude Skills, including how to avoid common mistakes and how to build “meta‑skills” that can be reused to construct other skills. - Skills act as plugins or extensions that give Claude specialized instructions while reducing prompt length; they can be loaded from local folders, uploaded as zip or the newer *.skill* files, or managed via the API (which requires code execution/file‑creation to be enabled). - A major source of confusion is that each platform expects a different packaging format: Claude Code looks for folders under */skills*, the web/desktop apps accept zip or *.skill* files, and uploaded zip files must be extracted into the correct local directory to be recognized. - Despite these quirks, the zip‑based skill format is interoperable—e.g., ChatGPT can open a *.skill* file as a zip—so you can leverage the same skill assets across different AI tools, provided you follow the proper extraction and placement steps. ## Sections - [00:00:00](https://www.youtube.com/watch?v=WKFFFumnzYI&t=0s) **Creating and Sharing Claude Skills** - A comprehensive tutorial that walks viewers through building Claude skills, avoiding common mistakes, leveraging meta‑skills, and handling the various file formats and deployment methods. - [00:03:20](https://www.youtube.com/watch?v=WKFFFumnzYI&t=200s) **Common Claude Skill Pitfalls** - The speaker outlines frequent errors such as forgetting to enable code execution, providing vague skill descriptions that misguide Claude, using unclear markdown headers, and suffering from inconsistent versioning, urging developers to be explicit and maintain a single source of truth. - [00:07:18](https://www.youtube.com/watch?v=WKFFFumnzYI&t=438s) **Building a PowerPoint Chunking Skill** - The speaker outlines how to create a skill that automatically checks and splits large PowerPoint requests to avoid exceeding Claude's context window, ensuring smoother generation. - [00:10:36](https://www.youtube.com/watch?v=WKFFFumnzYI&t=636s) **Automated Skill Generation Walkthrough** - The speaker walks through an 11-step automated process that creates a PowerPoint chunking skill, outputs documentation and strategy details, and packages everything into a functional but visually garbled zip file. - [00:13:44](https://www.youtube.com/watch?v=WKFFFumnzYI&t=824s) **Simplifying Skill Management for Non‑Technical Users** - The speaker describes how meta‑skills and built‑in tools such as profilers, gap analyzers, and token‑budget advisors streamline the creation, tracking, and optimization of AI skills, making complex work accessible to non‑technical people across platforms. - [00:17:10](https://www.youtube.com/watch?v=WKFFFumnzYI&t=1030s) **Building Prompt Skills for Testing & Security** - The speaker describes creating dedicated testing and security prompt skills, advises naming them after functions rather than teams, and previews a Claude continuous‑learning skill. ## Full Transcript
Claude skills took the world by storm
last week. This video is all about
giving you a complete tutorial so that
you know how to build a skill yourself
by the end of this video. But we're not
done there. We are not only going to
teach you how to build a skill. I'm
going to walk you through the common
pitfalls that I see. And I'm also going
to make sure that you understand the
kinds of skills that you can use to help
build other skills. I call them meta
skills. They aren't very widely in
circulation yet. I've built a bunch of
them for this video and I want to walk
through several of them with you so you
one know they exist and can grab them
and two know how to make them because I
think that's the more valuable piece. So
with that, let's get to it. First, the
big picture. Skills are a way to extend
what Claude can already do by giving it
very specialized instructions and tools.
They save a lot of weight on the prompt,
which is something I pointed out in my
first video right after Claude launched.
So, think of them like plugins or
extensions. You have one format and you
have three ways to use it.
Fundamentally, Claude code can read
skills directly from folders on your
computer. The web or desktop version
requires you to upload files through
settings. Traditionally, that has been
zip files. And by traditionally I mean
for the last week now they also can
accept a skill file type which is brand
new that Claude has introduced. I'll get
to that later. Finally, the API lets you
create and version skills
programmatically. One critical
requirement, you must enable code
execution or file creation in your
settings on the app or skills will not
work at all. And I saw people get
tripped up by that. Speaking of trip
ups, I have now seen a bunch of people's
skills in the wild. I've seen people in
my DMs asking me about skills since I
did my initial video. I want to give you
the top 10 things that have tripped
people up since Claude Skills launched
and how to address it. Number one, every
platform wants skills packaged
differently. And that is super annoying,
but it remains true. So, Claude Code
expects folders in specific locations
like /skills and web and desktop apps
will only accept zip files or skill
files, which by the way guys, if you're
wondering what a skill file is, it looks
like it is just a fancy extension on the
end of a zip. And I know that because I
threw a skill file into chat gpt and
guess what? It still works. Chat GPT can
crack it open if you tell it's a zip
file and can look inside and tell you
what's in there. So, you know that hack
I gave you last week where I said you
can use these skills in chat GBT chats?
Yeah, you can still do that. That still
works. If you upload a zip file to the
web version, Claude code will not see it
unless you extract it to the right
folder. This is another thing that trip
people up. Basically, Claude lives
locally on your computer, but also in
the web. And if you upload a skill, it
doesn't magically transfer to both
places. So, you have to think about
where you want the skill to be
accessible and do you need to put it in
a local folder? Do you need to upload it
to the web or the desktop so you can
chat with it in the interface? That's up
to you. I'm going to demonstrate the
interface because it's easier for people
to understand, but I wanted to give you
a sense of the breadth of Claude so you
understand why people get tripped up.
Trip number two, people forget to enable
code execution. I said this at the top
of the video, you got to enable code
execution. If it isn't enabled, you
can't upload skills and you can't run
them. Number three, skills don't trigger
when you expect them to. This one's
trickier, but Claude decides which skill
to use regardless of the surface. If
you're on Claude Code, if you're on the
web, whatever, it decides to use a skill
based on your description. So, a vague
description means Claude might not pick
your skill or might pick the wrong
skill. That means the first lines of
your skill.mmarkdown need to be
extremely specific about what it does.
Your skill.mmarkdown, by the way, is
just a plain text file that Claude can
help you build or you can build directly
that then gets zipped up into a zip
file, which Claude will accept, or this
fancy newskll ending on the file type,
which is basically the same as a zip,
but Claude decided to make it fancy.
Essentially, if that first line of text
isn't super clear, you're going to miss
out. And you'll see in the tutorial in a
second what I mean by super specific,
cuz I have an example. Number four,
versioning chaos matters. If you are
someone who uses claude code and the
interface and the app, you need to think
about which is true and real and how you
version these things. So pick a source
of truth and then put that skill into
every place you can think of if you're
one of those users that use all the
surfaces. Otherwise, you're going to get
confused. This brings me to number five,
which is there is a limit on the number
of skills you can access, at least in
the web and app versions. Claude limits
you to 20 uploaded skills and it does
enforce it. I have already hit that
limit and I have to make hard choices.
So think about the skills that you are
going to choose to invest in and make
sure you pick the right ones. I expect
that to shift as Claude gets more
capable over time, but that's the limit
we're at today. Next, you need to be
ready for a team level of confusion
around skills if you don't impose a
degree of order on common skills because
claude right now is just going to let
everybody upload everything in their own
cloud instances. And if you're in a
team's instance, that means that
everyone is going to have their own
version. There is no push to everyone
button for skills yet. I expect we'll
get there, but it doesn't exist. So,
everyone has their own individual
skills. And if you're an administrator
in a team's instance, if you're trying
to manage an enterprise install for
claude, that's something you'll have to
think about. That, by the way, is also
true if you are one of those people
who's doing the hack I mentioned in my
first video where you're using these in
chat GPT. Everyone in chat GPT is going
to be using these sort of skill things
individually
as a hack. And there isn't going to be a
teamwide instance unless you decide to
create a repository of skills that
everybody has to draw from. So think
about that if you are responsible for a
team. The next one I'm going to cover in
more detail. Security is not automatic.
And Claude does warn you. Skills may run
code. Skills may run scripts. You are
responsible for what you run and you
need to treat third party skills with
care to make sure that they're not
malicious. Last but not least,
documentation is super scattered right
now. I could not even find the updated
documentation on the skill file
structure. I just had to discover it and
tell you about it. There is this is
evolving so fast that we don't have
consistent documentation. So with that
in mind, look at the documentation you
can get from Anthropic and then also
look at other excellent sources of
documentation that are assembled from
the web. I'm going to link to a few in
my write up on the blog, but you got to
trust the community on this one to help
you assemble documentation. I don't
think Anthropic realized how big this
would be when they let it out. Okay, so
those are some of the pitfalls. We've
talked about the team side. We talked
about version chaos. We talked about
skills not triggering when you expect,
forgetting to enable code execution.
These are all things I have seen in
practice and I just wanted to get them
out of the way at the top because I
don't want you to get tripped up on the
value of skills and I want you to be
able to do something useful with them.
So with that in mind, let's jump into
how you actually build a skill. Okay,
here we are at the top of a tutorial on
building a skill. This is a PowerPoint
chunking skill. Remember how I promised
to solve some of these meta problems in
this video? Here's how I'm doing that. I
found in practice one of the biggest
issues that people had using skills is
it was tough to create powerpoints using
skills that did not run into Claude's
context window and generate a really
frustrating experience where you put all
the data and you put all the narrative
in and you get your fancy prompt and
then you put in your Claude skill and it
runs out of space. So I wanted to say is
there a skill I could build that Claude
could use to proactively help avoid that
situation and that's what this is all
about. So I my initial prompt is very
simple, right? There's not magic words
here. Help me build a PowerPoint chunker
skill. The idea is it is invoked when a
PowerPoint is asked for and it checks
how big the ask is. This is actually
important. I was very careful with that
wording invoked when a PowerPoint is
asked for because I want the skill to
invoke
so that any PowerPoint that's asked for
gets at least a look. And so then I say,
listen, I got to be honest with you. I
have some other skills in my in my
budget here. Are there other skills that
overlap with this? And so the first
thing Claus does is Claude goes and
looks and says, "Well, we have a token
budget adviser, which is another custom
scale I built that I'm going to stick on
the Substack. You have a pitch deck
builder and you have a general
PowerPoint builder, but Claude agrees
there doesn't appear to be a general
purpose chunker that triggers on any
PowerPoint request." So, it understood
my original prompt and it recommends
chunking for large presentations. So, it
does see an angle here that the three
skills up here do not cover, right?
There's the token budget, the pitch deck
builder, and the PowerPoint. And so then
it says,"I understand the challenge.
Would you like me to try and build for
this?" And I say, "Yep." And it just
starts building, right? It reads the
general PowerPoint skill to see if it
has chunking logic. It takes a few
seconds to look at gaps. And you can
actually pop this open and it will
actually give you like a lot of
specifics around how it's thinking about
each of these skills existing chunking
logic, right? Which is super helpful.
And you can dive in. I love the way
Claude shows this. When you look at the
three skills together, Claude says you
get this state, right? like what each
one does. The token and budget advisor
triggers on heavy input tasks, which is
what I built it to do. It can mention
presentations, but it's really designed
for something more, which is exactly
what I built it to do. Um, and then it
has the pitch deck builder, another one
I built, and it triggers during the
workflow once you're already in it. Um,
and so that's not quite right from a
chunking perspective. So Claude is
taking a deep dive. And one of the
things I love is that you can have
Claude do so much of the lifting. I
don't think people realize this. People
think they have to sit down and like
manually write a markdown file, and you
can, but you don't have to, right? You
don't have to. So, it agrees there's a
gap. It has a recommendation. It offers
to draft it, and it offers a sort of
distinctive approach. And I say, you
know what, to be honest, I think this
looks good. I don't really have any
concerns here. And so, it then goes and
executes 11 different steps to build it.
Let's just take a look at some of those
11 steps. It's going to architect the
skill. It reads the skill creator tool.
It creates the PowerPoint chunker skill.
It looks for packaging scripts. All of
this stuff it's doing in the background.
And eventually it comes out with three
documents. There's the skill itself with
that infamous.kill extension. Don't
worry, that's just azip in a disguise
for Halloween, I guess. Um, it comes out
with documentation because if anyone
knows Claude, Claude loves
documentation. Um, and it actually comes
out with a bonus document about how it
fits to existing skills. Uh, it then
tells you what it does. It re-emphasizes
my original request that it triggers on
any PowerPoint request. And then it
gives me the four strategies it's going
to use for chunking, which I love that I
get multiple strategies that will choose
between. So, sequential, structure,
first, etc. It explains why it's not
duplicative. And here are the files. And
I can look at them. I can say, "Oh,
well, let me check out the readme,
right?" Like, how does this actually
work? Super clear readme. I can read it.
I can understand. I can stick this
readme somewhere so I can follow it up
later. Can give me some examples. I love
this. It's great. And then you might
think, oh, I can then look at the zip.
Well, I got to tell you, the zip is not
going to render really pretty. It looks
like Dingbat's font in here. Don't
worry, it's still a functional skill.
That is just the way the zip is rendered
in. That's an example of how easy it is
to create a skill because all I have to
do is I just download that and then I
upload that into my capabilities section
and it will be right there. I'll show
you. So, here I am in my capabilities
section on Claude. I got there through
the settings section of Claude and I
went down and just clicked capabilities.
here. You might not see where it is
right away, but really all you have to
do is scroll down, right? You just
scroll down to skills, which is in
preview mode. You can upload a skill
here if you want. And you can also
enable or disable skills with this
toggle here. That works really easily.
If you click the dots, you can delete.
It's pretty self-explanatory. And
there's our new skill, PowerPoint
chunker. I just created it. Uh you can
see what it does very briefly. You can
expand the description so you see the
full thing. And you can It's just there.
It's active. and Claude will call it
when the time is right. And that's it.
Like, it's pretty simple. Now, you might
wonder what other skills you can build
that would help you with building
skills. What are meta skills that are
useful? Well, I have a collection here
that I've put together that I'm going to
share. Uh, I think the PowerPoint
chunker is super useful. I think the
skill security analyzer is a useful one.
So this one I built because I feel like
one of the gaps which I highlighted
earlier in this video is that Claude
does run code with this and we need some
kind of security analysis. So why not
use a skill to do it? I have a skill
debugging assistant. So when people have
trigger failures or parameter problems
or prompt conflicts, how can we start to
address that? Um I have a documentation
generator for skills in case people
forget their documentation. I have a
testing framework that provides test
cases for skills so that you can
actually test them. I have a dependency
mapper. By the way, if this starts to
look a little bit like software, this
goes back to one of the things I've been
calling out with prompts as a whole,
which I consider skills a subset of
prompts. You're feeding the machine
context to work for you. You got to
treat it like code. You're giving the
machine inputs. So, you treat prompts
like code, you treat skills like code.
The nice thing about skills is it has
never been easier to do that for a
non-technical person because here it is.
It will not change on you, right? like
you write the skill, you upload it, you
get it, and it's just there. You don't
have to remember it. And that's why I
keep emphasizing this is a substantially
easier way to do work and to do complex
work than we did before because it makes
a lot of the steady context you need
something you don't have to worry about.
It just sits here in skills. What I'm
aiming to do with these meta skills is
to take the steady context of we have to
check for security, we have to debug it,
we have to have a testing framework, we
have to have dependency mapper. Like I'm
assuming because of the response that
I've seen that people are starting to
build dozens and dozens of skills and
they're going to need tools like this to
help them track them and handle them. I
even have like a performance profiler,
an diagnostic tool for analyzing and
optimizing skill prompts, a skill gap
analyzer if you're wondering what you
can use, and of course, everybody's
favorite, the token budget adviser. It
tells you if you are going to run out of
tokens. So, my goal here is pretty
simple. This is becoming a big part of
the ecosystem. There are hacks that take
this right away into chat GPT. Like I
said, you can just upload the skill
file. And so, this is not going to be
limited to Claude for very long. it
already isn't if you want to be
adventurous. How do we start to go from
what a cool toy to we can do real work
with this? And I think so much of that
comes down to having useful
infrastructure like this that enables us
to take skills seriously and actually
build with them. All right, so we've
done a bit of a tutorial. I'm going to
give you one more peek at a really cool
skill before the end of this video. But
next, I want to just get into some best
practices that we're starting to see
emerge after the first week or so. And
we've had a lot of people sort of beat
on cloud skills and see what works. I
think that the first thing I want to
call out is in line with that idea of
treating skills like code, but I want to
make it accessible, right? You can
decide how seriously you want to take
skills. If you just want to yolo your
way through and throw a skill in, it
will work. If you feel like you need
skills to work for serious work all the
time, I would encourage you to use
version numbers. I would encourage you
to keep a change log. I would encourage
you to store your skills in some
dedicated place with a structure so that
you can co go back to them and find them
again. I would also encourage you to
design for discoverability in your skill
markdown file. Make sure that the first
paragraph is super clear. Use this skill
when you need to define specific tasks.
Right? It takes specific inputs. It
produces specific outputs. Don't use it
for things you don't want it to trigger
for. that that is one of the best hacks
you can have because it takes away the
missed trigger issue that people are
having with Claude skills where the
markdown starts with principles or
something vague and Claude progressively
reads the skill and just doesn't get to
the trigger in time. The other thing
that I would call out that is popping
from a best practice perspective is the
more you can have one workflow for all
the places where you use skills, the
better. So for cloud code, for the web
and the desktop, for the API, have the
same skill in a folder somewhere and
just like copy it into the web and
desktop, copy it to the right folder for
code. Just make sure it's not going to
be an issue where you have like
similarly named skills in different
versions on different surfaces. If
that's going to be you, like a little
organization is going to go a long way.
Testing before you ship matters a lot.
That's why I created a prompt uh skill
to help with that. Right? there's
literally a skill I'm shipping to help
you with testing because I think that's
something that's really annoying to do
but it needs to be done and I thought
why not write a skill for it. And then
for the security side, I also wrote a
skill to sort of assess the skills you
may create for security. And it goes
after standard vulnerabilities, right?
Like are there issues here with uh
dependencies we don't trust? Are there
issues here with uh code running in ways
that would perhaps not be appropriate
for a clawed web app to run, etc., etc.
There's a lot that goes into the
security side. I would encourage you
also, and this is just a small tip, but
if you're in a team environment, name
your skills after jobs, not teams. I
know that sounds weird, but the more you
can be clear about what the skill does,
the more you're likely to keep track of
it and use it for that purpose as
opposed to naming it like the product
management team skill for whatever,
right? Like just drop the team names,
drop your name out of it if you can, and
just name the thing for what it does.
And I think that's going to be useful
for you. Okay, before I go farther, let
me just show you a really cool skill
that falls in the meta-kill category,
but I don't think anyone's doing this.
This is the Claude skill for continuous
learning. Okay, here's the idea. I want
to take this idea that I've been sort of
finding across the web around continual
learning as a challenge. What would it
look like? And I want to turn it into a
skill because no one's done that and I
think it's super interesting. And so I
take that as a challenge and I say
within the constraints that you've got
tell me how you would solve this problem
and I want to make it clear right think
about your current constraints don't be
aspiring. Okay let me think about what's
actually possible says Claude. And so
Claude basically identifies that the key
thing the bottleneck is manually adding
any new update. But Claude could
identify opportunities to create new
skills autonomously within natural
conversation which would be a huge plus.
Claude can recognize learning moments.
It can document the learning in a
structured format and recommend a skill.
And it's sort of gives me a sample of
how this would work. Um, and then it
starts to sort of get into how Claude
and I would have to partner to do this
because if you're going to do continual
learning with an AI, you have to have a
contract or agreement between you and
the AI as to what you're going to learn.
And so I say, "Okay, I can live with
uploading this. What if we proceed to
write this as a a skill? What would that
look like and so it just starts to dig
in, right? Um and it says, "Let's look
at these sort of issues. Would we want
to tackle novel problem solving,
repeated patterns, domain specific?" And
you notice that Claude is getting
excited about this. This is going way
beyond any seed of text that I found.
And it's getting into
how you would actually operationalize
this. And so then I kind of read through
and I give my opinion and I say, "This
is the goal that I have. I want you to
start at being able to identify 80% of
tasks that hit the context window before
10% of the response tokens are used.
Clause like thank you for the clear
constraint. You're basically asking when
you have complex challenging pieces of
work, how do ident how do we identify
that in a way that's useful and generate
a skill opportunity something we can
learn and drive? And then Claude says,
"Okay, this is great." Uh, and starts to
build. And long story short, I can build
a continual learning capture skill and
stick it into Claude, which I think is
super cool. So there you go. We've
gotten through two different skill
tutorials here. You've seen how it
works. I've given you some of the
pitfalls that have caused people to trip
up on Claude skills and I've given you
some best practices. If we zoom out the
camera lens just a minute from all of
this tactical stuff. The reason why I
think this matters is because Claude's
skills are one of the handiest ways to
extend our prompting power that I have
ever seen. It is so easy now to do
harder pieces of work because you can
write one skill for all of the stuff
that you don't want to just repeat
yourself on over and over and over
again. And that's why people have gotten
excited about this. That's why it is
worth it to dig in and learn this. And I
don't want you to lose the value there
just because you have to wait into some
of the detail here. This is absolutely
worth it. So, I hope you enjoy the Cloud
Skills tutorial. If you want to go and
grab all those skills I demoed, I have
them over on the Substack. And honestly,
have fun. Like, this is an amazing
moment in AI because we all get to build
these skills and talk about them and
learn about them. And uh best of luck in
the wild new world of cloud skills.