The argument in one line.
The gap between a skill that breaks down after a week and one that compounds over years is three deliberate additions: a verification layer with an objective pass/fail output, a gotchas section grown from real failures, and a description field written as a trigger condition rather than a summary.
Read if. Skip if.
READ IF YOU ARE…
- You have built a few Claude Code skills but they feel brittle or produce inconsistent output after a week of real use.
- You are a non-technical operator in marketing, legal, or growth wondering whether skills can apply to your workflows.
- You want the conceptual model behind why skills work rather than just another template to copy.
- You are burning tokens on repeated boilerplate and have not yet separated deterministic steps into scripts.
SKIP IF…
- You want a live code walkthrough of a complete skill file — the video stays conceptual with on-screen prompts.
- You have already read the Anthropic skills best-practices blog post — there is significant content overlap.
The full version, fast.
Anthropic maps nine technical skill categories to four practical types: utility, verification, data enrichment, and orchestration. A skill is a folder, not a file: scripts handle deterministic steps, assets lock the output format, and setup prompts make it human-readable years later. Verification is the highest-ROI investment — two kinds exist (correctness and quality) and both need an objective pass/fail or graded output. The gotchas section turns a day-one draft into a compounding moat, but only by logging real Claude failures, never hypothetical ones. Finally, the skill description field is a trigger condition naming who it serves and the exact words a user types — not a summary of what the skill does.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:21
01 · Cold open
Anthropic released their internal skills playbook; this video extracts five lessons anyone can use today.
00:21 – 03:02
02 · Lesson 1 — Understand the Categories
Nine Anthropic technical categories collapse to four skill types. Best skills fit cleanly into one.
03:02 – 06:55
03 · Lesson 2 — Use the Power Components
A skill is a folder, not a file. Three components: scripts (deterministic steps), assets/templates (lock output format), setup prompts (config.json, ask-user-question, declared arguments).
06:55 – 11:21
04 · Lesson 3 — Focus on the Verifier
Verification is Anthropic's highest-impact skill type. Two kinds: correctness and quality. Requires objective output. Amol Avesar's manager-simulator skill is the standout example.
11:21 – 13:30
05 · Lesson 4 — Write the Gotchas
Living log of real Claude failures added to the skill over time. Do not prefill with hypothetical edge cases.
13:30 – 15:52
06 · Lesson 5 — Tune the Trigger
The description field is a trigger condition, not a summary. Orchestration skills chain named sub-skills.
15:52 – 16:15
07 · Outro
Five-lesson recap. Pointer to next video on what is worth building.
Lines worth screenshotting.
- Verification skills have had the most measurable impact on Claude output quality internally — more than any other skill type.
- A skill is a folder, not a markdown file: the scripts, assets, and data inside it are what make it powerful.
- Giving Claude a script to handle deterministic steps lets it spend its turns on composition rather than reconstructing boilerplate.
- There are two kinds of verification: correctness (did Claude get the facts right?) and quality (does the output meet your bar?). Most people only think about one.
- The description field in a skill's frontmatter is a trigger condition, not a summary. Claude scans it to decide whether to fire the skill.
- A good trigger condition does two things: names who it serves and specifies when it should fire — in the exact words a user would type.
- The gotchas section should only contain things Claude actually did wrong. If Claude has not done it yet, you do not need a gotcha for it yet.
- Orchestration skills should be derived from sub-skills, not built from scratch. Update a utility sub-skill once and every orchestration that calls it updates automatically.
- AI is most valuable as an amplifier (same quantity, higher quality) not a multiplier (more output, same quality).
- A config.json file lets a skill ask for missing values on first run and remember them — eliminating repeated setup prompts in future sessions.
- Skills are living documents: most of Anthropic's best skills started with a few lines and a single gotcha, then improved as Claude hit new edge cases.
- Encoding a specific person's judgment as a verification skill — from their writing and conversations — lets you pre-clear work against their standard before they ever see it.
Five rules that determine if a skill compounds or collapses.
WHAT TO LEARN
A skill built without a verifier, a gotchas log, and a trigger condition will drift and break; a skill built with all three improves automatically the more it is used.
- Every skill should fit cleanly into one of four types: utility, verification, data enrichment, or orchestration. Skills that straddle types confuse the agent and produce inconsistent results.
- A skill is a folder, not a markdown file. Scripts handle the deterministic steps so Claude spends its turns on judgment, not boilerplate.
- Assets and templates lock the output format. When you change the template, the skill's output changes automatically without touching the prompt.
- Verification is the highest-return investment in any skill system. Design verifiers around objective outputs: pass/fail or a grade out of 10, never a vague qualitative impression.
- There are two distinct verification types. Correctness checks whether facts and numbers are right. Quality checks whether the output clears the bar you care about. Both are necessary.
- Encoding a specific person's judgment as a verification skill — from their writing and recorded feedback — lets you pre-clear work against their standard before they ever see it.
- The gotchas section is a log of real failures, not a speculative list. Only add a gotcha after Claude actually hits the edge case. Hypothetical gotchas dilute the signal.
- The description field in a skill's frontmatter is a trigger condition, not a summary. Write the exact natural-language phrases a user would type when they need the skill.
- Orchestration skills should chain existing sub-skills by name, not rebuild their logic. Update a sub-skill once and every orchestration that calls it updates automatically.
Terms worth knowing.
- Utility skill
- A small, focused skill that does exactly one task. Often used as a building block inside larger orchestration skills.
- Verification skill
- A skill that checks the final output of another skill or workflow. Requires an objective output: a clear pass/fail, a grade out of 10, or a named verdict.
- Data enrichment skill
- A skill that pulls external data into the system to improve the final product — for example, fetching website analytics, competitor reports, or API data.
- Orchestration skill
- A skill that chains other skills together into a multi-step workflow. Should always be derived from pre-existing sub-skills rather than rebuilt from scratch.
- Deterministic workflow
- A step where the same input always produces the same output. Best handled by a script, not by AI.
- Nondeterministic workflow
- A step where the same input can produce different outputs — the appropriate domain for Claude's turns.
- Gotcha
- A documented failure mode added to a skill's SKILL.md file after Claude actually hits it. The highest-signal content in any mature skill.
- Trigger condition
- The description field in a skill's frontmatter, written to tell Claude when to fire the skill rather than what the skill does. Should include the exact natural-language phrases a user would type.
- Skill-driven verification
- The practice of adding a verification component to an existing skill so that it returns a clear pass/fail or grade — rather than building a separate standalone verifier.
Things they pointed at.
Lines you could clip.
07:00
“Verification skills have had the most measurable impact on Claude's output quality internally.”
07:12
“It can be worth having an engineer spend a week just making your verification skills excellent.”
07:34
“99% of people think about AI as a multiplier. Higher quantity, same quality. But I found it's most valuable as an amplifier. The same quantity, but a higher quality.”
12:41
“Think of the skill as the structure, a verifier as the leverage, and the gotchas as your personal moat.”
13:54
“The description field is not a summary. It's a description of when to trigger the skill.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphoranalogystory
00:00Anthropic, the inventor of Claude, just released their internal playbook on how they use Claude skills. Their engineers, their marketing team, their legal team, they've all streamlined their technical and nontechnical workflows with skills. And after analyzing everything their team has posted from interviews to blog posts to official docs, I've broken down exactly how they use Claude skills into five simple lessons anyone can follow and implement today.
00:22Lesson number one is understand the categories. The Anthropic team broke skills down into nine categories which you can see on the screen. Each of these are technical because they are made by the creators Claude Code, but the reality is that all nine of these technical categories can be bucketed into four different skill types.
00:37And these skill types can apply to both technical and nontechnical people. So skill type number one is utility skills. These are small reasonable things that do one specific task and they're often layered with other larger skills.
00:49So a technical utility skill, this could be how you interact with specific libraries or APIs. And a nontechnical example could be draft response in my voice or simplify writing. Skill type number two is verification skills.
01:01These skills check the final output of the product. Pulled directly from their blog post, verification skills have had the most measurable impact on Claude's output quality internally. So these are hyper critical, so we'll dive into them more in a later lesson.
01:13But for a technical use case, this could be product verification or code quality review. For a nontechnical example, this could be something like brand voice check or style guide review. Skill type number three is data enrichment skills.
01:25This pulls external data into your system, which will help enhance the final product. A technical use case could pull website performance data or conversion traffic from a website. For me, I do something very similar.
01:35I have a skill called funnel digest, which looks across all my websites and analyzes the traffic. A non technical use case in this example could analyze competitors, pull recent consumer reports, etcetera.
01:45That's the first three skill types, but the fourth is an orchestration skill. These are skills that are specifically designed to chain steps and skills together. An example of what an orchestration could look like is, let's say, there's an orchestration skill called generate report.
01:58This would pull website analytics, which could be a data enrichment skill. It'll then draft a write up, which could be a utility skill. And then it'll verify branding and numbers, which could be a verification skill.
02:07All of these orchestrated together to create one final report. So one command, which stacks all of the other buckets together. Now I'll be going through this more in lesson five because you have to be very careful when you're doing this.
02:18So those are the four different categories, and this will help you think about what types of skills you can actually create. But as you're creating these skills, you have to avoid this one specific trap. The best skills fit cleanly into one category.
02:29The ones that try and do too much straddle several of these categories and confuse the agent. So you want specific skills that fall into a specific skill type. With this in mind and taking these four categories, here is a prompt that you can run that looks at your existing skills along with your past chat history to identify potential skills and what bucket they fall into, and if any should be adjusted so that they fit clearly in a specific bucket.
02:51Now that's about how Anthropic thinks about the categories of the different skills that you can create. But once you actually decide the skills that you wanna create, there are some key features that you have to include. Lesson number two is use the power components.
03:04Before you build any skill, you have to know what a skill is conceptually. From Anthropic, a common misconception we hear about skills is that they are just markdown files. But the most interesting part of skills is that they're not just text files.
03:15They're folders that can include scripts, assets, data, etcetera that the agent can discover, explore, and manipulate. To summarize, a skill is a folder with a bunch of components inside of it, and you need to understand what these components are so you can optimize these skills. And there are three components that I think are the most important to get right.
03:32The first is scripts. So a script is computer code that runs to complete a task. And from Anthropic, one of the most powerful tools you can give Claude is code.
03:39Giving Claude scripts and libraries lets Claude spend its turns on composition deciding what to do next rather than reconstructing boilerplate. By creating a script, you're allowing Claude to focus on new things instead of things it's already figured out.
03:53And by creating scripts, you're able to partition deterministic and nondeterministic workflows.
03:57A deterministic workflow is where the same input will produce the same output. So two plus two always equals four. It's something that's repeatable and a script handles this perfectly.
04:06A nondeterministic workflow is it could have the same input, but it has different possible outputs. And this is what you've seen when you're using AI. The more you push into scripts, the more repeatable your output will be and the fewer tokens you'll burn.
04:18So you wanna try and have your skills use code to run the deterministic parts and use AI for the nondeterministic parts. Don't worry, I'm gonna give you a prompt to help you with all of this at the end of these three points. The second power feature is assets slash templates.
04:30Within a Claude skill, you can upload specific files as templates for the output that you want to be created. For example, if your target output is a PowerPoint presentation, you can upload a template of the past report which will live in your assets folder.
04:43Cloud will then use this as a starting point to fill in for all future reports. And that way if you change the template, the skill will automatically change to the final output. Similar to scripts, this is about taking your skill and making it so it stops improvising and starts doing the same thing every single time.
04:58The third is setup prompts. When building skills, unfortunately, you will quickly forget how the skill works. So it is critical that you make them clear and easy to use.
05:06There's a famous saying, any fool can write code that a computer can understand. Good programmers write code that humans can understand. Translating this to the AI world, anyone can write a skill that Claude understands.
05:17The best people write skills that humans understand too. So how do you do that? Well, there's three things to do when you're setting up your skills.
05:23Create a config dot JSON file so that on the first run, if a value is missing, the skill asks for that specific information. Then once you provide it, it'll then be stored so you don't have to provide it in future runs. It's a way to set up the skill so that it remembers tasks going forward.
05:37The next component is ask user questions. The next thing you wanna do is use ask user question tool. For multiple choice setups, you can tell Claude to use ask user question tool inside the skill dot m d file.
05:48Claude will then present a structured option instead of free form text. It's similar to the interview flow that you may have seen before on Claude, but this is how you enforce it within your skills. Here you can see me using it for one of my skills where it's directly prompting me to answer questions and it's just a very user friendly experience.
06:04And the last thing that'll help these skills be more readable in the future is setting up arguments. At the top of a skill.md file, you can declare an arguments field. When someone calls a specific skill, this will serve as a reminder to whoever using that skill for what input is needed for it to run properly.
06:20And for all these, don't think about yourself today. Think about yourself one, two, three years from now. Take your time and build for the future.
06:26That's who we wanna think about. And if you've been watching this channel, you know that I believe in building sustainable systems that compound over time. I can't stress how important this is.
06:36So here's a prompt to audit every skill in your project and identify which of these three components they should be using to enhance these skills. So now you know what's possible inside the folder itself.
06:46There's one bucket I touched on earlier that we need to dive deeper on, and this is the one that Anthropic ranks as the most important bucket in your entire skill categories. Lesson number three is focus on the verifier.
06:57From Anthropic, verification skills have had the most measurable impact on Claude's output quality internally. It can be worth having an engineer spend a week just making your verification skills excellent. So clearly, verification is important.
07:09And even the creator of Claude code has said if you give Claude a way to verify, it will two to three x the quality of the output. So clearly, it's important, but most people miss this important fact.
07:19There are actually two types of verification you need to understand. The first is verify for correctness. Did Claude get the facts right?
07:26Are the numbers correct? Or is the app running? Like, did it do what it needed to do?
07:30This is the more quantifiable outputs. But the second is verify for quality. Does the output actually meet the bar you want?
07:3799% of people think about AI as a multiplier. Higher quantity, same quality.
07:42But personally, I found that it's most valuable as an amplifier. The same quantity, but a higher quality.
07:47And by treating Claude's skills as a quality verification layer, you raise the bar on every output you produce. So both of these verification types are important. So how can you set them up?
07:57Well, right out of the box, Anthropic already has two skills built into Claude code. Slash verify, which will run your app and confirm a code change actually does what it's supposed to. And slash run launches the app so that Claude can see its own work.
08:09These are both technical examples and you have them by default, which is nice. But what if you wanna make your own verification skills? The key thing that any good verifier needs is an objective output.
08:18And this is something that either you or Claude can tell if the test passed or not. For example, if you have a slash code reviewer skill, you want it to say pass or fail. Or if you have a report review skill, you want it to give a grade out of 10.
08:30So you can build these verifiers from scratch, but I prefer doing something different. I prefer altering existing skills so that there's a clear verification component. I call this skill driven verification.
08:40For an example, let's say you have a brand voice skill. You could add a verification component to it so that you could run slash brand voice and will tell you if it passes or fails the verification. You're essentially tweaking existing skills to have an output that's easier to verify.
08:55To do this, here are two prompts. The first audits your existing skills to see which ones can be tweaked into a verifier. The second helps you build a verifier from scratch and identifies any external tools you need to pull in data to verify the output.
09:09So the prompts are on screen to help with verification. And when studying how Anthropic uses skills, there was one example that stands out the most to me about how to verify outputs. Amol Avasar, their head of growth, built a verification skill that simulates feedback from his actual manager.
09:23He fed Claude data from her public writing or her Slack conversations. And every week, the skill fires and tells him what feedback she would have given him. It's essentially her judgment encoded as a reviewer.
09:33Here's him in an interview talking about how you'd use a reviewer. I do that for myself as well. So I basically
09:39you know, one of my manager, Ami Vora, was, I think, a podcast guest of yours. Right?
09:44So I say, hey. Based on what you know of Ami both publicly, she's written extensively about product and then internally, and then our discussions,
09:53what what is every based on everything that I've done or not done this week, what feedback do you have for me as army? And, like, I get that every week. Right?
10:01So, like And so by the time he shows the manager anything, it's already passed her AI clones quality bar once. For me, do something very similar. I have a skill called internal focus group, and it runs through a group of advisers who will give me specific feedback on what I'm working on.
10:14As a founder, I don't really have a manager, but these serve as my board of advisers essentially. This has helped me so much that I actually built buildpartner.ai, which has over a thousand users to help people with exactly this.
10:26You can just call slash expert advice, and we'll give you feedback from experts on whatever you're working on. So at this point, you understand some of the most important lessons to use with skills. But there's one component that I recently learned about that I started implementing in every skill that I make.
10:38But before we get to that in lesson four, if this is your first video of mine, welcome to the channel. But if this is your second or more, here is our anti slob agreement. The visuals that testing the hours of research that went into this video, this is entirely built for humans, not for AI clanker robot scrapers.
10:53So all that I ask is that you subscribe as part of this agreement to help this content reach more people. Also, every video, I give away a Clawdmax subscription, so comment below with what you're building to enter. This video's winner is the GOAT, Matthew Hinman, who's building a Monday morning report generator.
11:08Also, do wanna shout out everyone. We recently got to 50 k subscribers. Thank you for everyone who watches and supports.
11:14I've been making videos for over six years and I just recently started getting traction. So I absolutely love seeing you guys in the comments saying how helpful these videos are. Anyway, enough of that.
11:22Let's go to lesson four, which is write the gotchas. This is a running list of issues Claude hits when using a specific skill. Think of this like a list of things not to do within a Skill MD file.
11:31From a recent blog post, the highest signal content in any skill is the gotcha section. These sections should be built up from common failure points that Claude runs into when using your skill. On the screen, there are three examples of gotchas that the Claude code team has in their skills.
11:46This is just an example of how when you're using skills, you'll realize some of the issues. So it's clear what it is. Right?
11:51It's a list of things not to do. But there is a meta point that makes this so valuable. At the end of the day, these are your skills, and these skills are living, breathing documents.
12:00The reality is you're not gonna get it right the first time. So these edge cases, you just haven't lived through yet. And the more that you use the skill, the more gotchas that you'll add.
12:08And just the fact of adding gotchas becomes a forcing function to actually use the skill and test it. Here's them talking about exactly this. Most of our best skills began with a few lines and a single gotcha, then got better because people kept adding to them as Claude hit new edge cases.
12:23Here's a graphic to show the visualization of what a skill could look like month one, month two, and month three. So if your skill isn't perfect on day one, don't worry about that because these skills should improve over time, and the gotchas are one of the most effective ways to do that.
12:36Think of the skill as the structure, a verifier as the leverage, and . And you may be thinking, oh, I'll just add a list of all the gotchas to start. Well, it's super important not to try and speed this process up.
12:47You should really only include a log of things that you actually saw go wrong. If Claude hasn't done it yet, you don't need a gotcha for it yet. Wait, watch, and then add to it.
12:56So to implement these gotchas into your skills, here is a prompt. And a personal preference of mine, I like formatting the output files of some of these reveals so that I can approve or deny the changes in a single file. So I'll have it create a markdown file formatted for Obsidian that looks like this.
13:09Here's a recent gotchas audit I ran, and I prefer just clicking through and signing off on the specific gotchas that I actually wanna implement. The reason being, I don't want any gotchas that aren't actually real gotchas in this part of the file. So we've learned how Anthropic uses skills, how to optimize them, and then configure them with gotchas.
13:26But how do you actually call them and how do you make sure that they're being used in your day to day? Lesson number five is tune the trigger. A trigger is what invokes the skill.
13:34This is how it's actually ran. Let's say you wanted to be the trigger. When you type slash draft email into Claude and then hit send, you're triggering the skill.
13:41But that's manual. Right? And we'd rather Claude automatically decide when to run it on its own.
13:46Technically speaking, here is how Claude actually decides when to use a skill. When Claude code starts a session, it builds a list of every available skill with its This list is what Claude scans to decide, is there a skill for this request?
13:57Which means the description field is not a summary. It's a description of when to trigger the skill. When I first read that, I was extremely surprised because it's not what you would normally think it is.
14:07The description should describe what triggers it. And a good trigger condition does two things. It names who it serves and when it should fire.
14:14Pulling directly from one of Anthropic's most famous skills, front end design, here is the description. Create distinctive production grade front end interfaces with high design quality.
14:23Use a skill when the user asked to build web components, pages, or applications. Generates creative polished code that avoids generic AI aesthetics. I won't read the whole thing, but the key thing here to notice is use this skill when that user asked to build, and then it lists typical human language that you would expect a person to say.
14:39This makes it super easy for Claude to know when to use the skill. So now if I type in create me a web page for this dashboard, Claude will automatically know to trigger the front end design skill. So when you're creating a skill, be explicit about what it does and think about the actual words you type when invoking it.
14:54And then those are the words that Claude needs to see in the description. So that's how you can get Claude to call a skill. But can skills call other skills?
15:02This goes into a skill type I mentioned earlier called orchestration skills. These string multiple skills together. And to make these, it's very simple.
15:08From Anthropic, you can just reference other skills by name, and the model will invoke them if they are installed. But the key with any good orchestration skill is that they should be derived from other skills. If you're building a complex workflow from scratch, ask yourself, can I break this into sub utility skills?
15:24Because these utility skills can be leveraged across many orchestration skills. So that way you don't have things overlapping reinventing the wheel. And another benefit is if you ever update the utility skill, the orchestration skill that calls it will automatically update as well.
15:38This makes the entire system so much more maintainable. So here's a prompt that will audit your skills and suggest improvements on everything that we covered in this lesson. At this point, you understand the categories of skills and how to bucket them, the key components to optimize, how to build verifiers, configuring gotchas to create your moat, and building out the right triggers.
15:56Each of these five lessons have been battle tested by myself as well as the Anthropic team. So make sure you follow everything precisely, and if you have to, go and rewatch the lessons I covered. So if you like this video, you will love this video where I dive into how the founders of Claude determine what is worth building.
16:09It'll give you actual advice on the things that you should and shouldn't use Claude for. I'll see you over there. Peace.
The Hook
The bait, then the rug-pull.
When Anthropic publishes its internal playbook, most viewers skim the blog post. This video does the work of reading every interview, doc, and post and collapsing it into five lessons — the kind you can act on the same day you watch.
Named ideas worth stealing.
00:21list
The Four Skill Types
- Utility Skills
- Verification Skills
- Data Enrichment Skills
- Orchestration Skills
Anthropic's nine technical categories collapse to four practical types. Key rule: every skill should fit cleanly into exactly one type.
Steal forAuditing an existing skill library to identify skills that straddle categories and confuse the agent
03:02list
The Three Power Components
- Scripts (deterministic steps)
- Assets / Templates (output lock)
- Setup Prompts (config.json, ask-user-question, arguments)
What goes inside the skill folder. Scripts partition deterministic from nondeterministic work. Assets lock the output format. Setup prompts make the skill human-readable years later.
Steal forAuditing any existing skill to identify which components it is missing
07:03list
Two Types of Verification
- Verify for Correctness (facts, numbers, sources real?)
- Verify for Quality (does output meet your bar?)
Both require an objective output a human or Claude can judge without ambiguity — pass/fail or a grade out of 10.
Steal forDesigning the output contract for any new verification skill
12:41concept
Skill as Structure, Verifier as Leverage, Gotchas as Moat
Mental model for how three layers of a mature skill relate. The structure is the folder and its components. The verifier is what multiplies quality. The gotchas accumulate competitive advantage that cannot be cloned without running the same failures.
Steal forExplaining the compounding value of skills to a team that only sees them as fancy prompts
How they asked for the click.
VERBAL ASK
10:30product
“You can just call /expert-advice and it will give you feedback from experts on whatever you're working on.”
Soft mid-video product mention for buildpartner.ai, earned by demoing the internal focus group concept first. Hard subscribe CTA at ~10:50 with Claude Max subscription giveaway.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
Visual structure at a glance.
