You know those books on your shelf? The ones you bought in a burst of motivation, read cover to cover thinking “this is brilliant,” and then… put right back on the shelf. Three months later, you trip over the exact same problem. The answer is 30 centimeters away on your bookshelf, but it can’t help you — because you’ve forgotten what Chapter 7, Section 3 said.

This is probably the saddest distance between humans and knowledge. You bought the book. The book didn’t use you.

Sahil Lavingia — founder and CEO of Gumroad — took a very different approach to this problem. He didn’t write another book, launch another course, or record a YouTube series to “help you review.” He took his bestseller The Minimalist Entrepreneur and turned it into 10 Claude Code skills.

A business book. Turned into slash commands. When you need it, you type one line, and the book’s wisdom comes to find you.

What Are Claude Code Skills: A Startup Advisor Living in Your Terminal

Quick context for those not deep in the Claude Code ecosystem. Claude Code has a skill system — think of it like apps for your phone. You package domain-specific knowledge and workflows into a /command that you invoke when needed. These aren’t generic “write me an email” conversations. They’re tool modules with specific trigger conditions, structured inputs, and structured outputs.

Imagine you have a friend who’s an incredibly experienced startup advisor. But they’re not sitting next to you waiting for questions — they live in your terminal. You call their name and they show up, remembering every point from every chapter of The Minimalist Entrepreneur.

Sahil essentially turned himself into this friend (the AI version).

Clawd Clawd OS：

“Turning your brain into installable software” — sounds like cyberpunk fiction, but Sahil actually did it. And here’s the ironic part: by doing this, he’s making his own book unnecessary to read. If /validate-idea works well enough, who’s going to flip to Chapter 4? But I think Sahil knows this trade-off — your methodology living in thousands of people’s terminals is worth infinitely more than gathering dust on tens of thousands of bookshelves (◍•ᴗ•◍)

Ten Skills: One Book, Ten Keys

The ten skills follow the book’s narrative arc, but they’re not just chapter summaries pasted onto a command — they’re more like Sahil’s decade-plus of startup experience compressed into individual decision engines. Let me walk you through them, and I promise it won’t read like a README.

The story starts with /find-community. Most people start a business by dreaming up a genius idea, then scrambling to find buyers. Sahil flips this completely: find your people first, and let the idea grow from their pain points. Sounds counterintuitive? But that’s exactly how Gumroad was born — Sahil was hanging out in creator communities, noticed they struggled to sell things, and built Gumroad. He didn’t start with “I want to build a payments platform.”

Next come two safety mechanisms: /validate-idea stops you before you write a single line of code to ask “wait, is this problem actually worth solving?”, and /mvp helps you control scope — note the design target is “ship in a weekend.” One. Weekend. If that feels impossible, your scope is too big, and that itself is a signal.

Then comes the most ruthless one — /processize. Before you automate anything, deliver the service by hand. No code, no servers — just email, Google Sheets, and manual notifications, serving customers like a human API. Sounds stupid? But it forces you to face a brutal truth: if you can’t do it well by hand, writing code just makes you do the wrong thing faster.

/first-customers — find your first 100 customers. Not 10,000 — one hundred. Because if you can’t convince 100 people face to face, throwing a million dollars at Facebook ads won’t help either — you’re just amplifying something that doesn’t work yet.

That’s the “survival phase” done. The rest — /pricing (charge from day one; free things have no feedback loop), /marketing-plan (make fans, not headlines), /grow-sustainably (spend less than you make — sounds obvious, but ask the founders who burned through everything before figuring that out), /company-values (build the house you want to live in) — these teach you “how not to die.”

The final key, /minimalist-review, is the Swiss Army knife — throw any business decision at it and let Sahil’s framework X-ray it for you.

Clawd Clawd 想補充：

The tweet says 9 skills, but the GitHub repo has 10 — there’s an extra /processize. My guess is Sahil finished, looked back, and thought “no, this idea is too important to merge into another skill.” How many indie hacker tombstones read “spent three months building before validating”? If you remember just one thing from this article, make it processize — do it by hand first, then decide if you need code (◍˃̶ᗜ˂̶◍)⁠ノ”

Prompts vs Skills: The Gap Between a Post-it Note and a Live-in Coach

Two lines to install in Claude Code:

/plugin marketplace add slavingia/skills
/plugin install minimalist-entrepreneur

Installation tutorial done. But what I actually want to talk about is why this is a completely different species from those “100 KILLER ChatGPT Prompts!” threads on Twitter.

Think about it. A prompt is a Post-it note — you write “remember to exercise” and stick it on the fridge. By day three, you don’t even see it anymore. A skill is more like a personal trainer living in your house — you don’t need to “remember,” because when you wake up, they’re already there, and they know what you need to work on today.

But the more lethal difference is this: these ten skills have causal relationships with each other. You wouldn’t run /grow-sustainably when you don’t even have a single customer — that’s trying to fly before you can walk. Prompt collections never tell you this sequencing, because they’re just Post-it notes scattered across a table. Each one has solid advice, but nobody tells you which one to read first.

And since skills live directly in your development environment, there’s zero friction. Stuck on a “should I build this feature or not” decision while coding? No switching apps, no flipping through books — just /minimalist-review. The book’s framework arrives in three seconds.

Clawd Clawd 吐槽時間：

You might think “aren’t skills just upgraded prompts?” Well, isn’t the iPhone just an upgraded iPod? Technically yes, but a qualitative shift has happened. You use a prompt once and toss it. A skill has trigger conditions, structured outputs, can be installed by others, and can chain with other skills. This is the leap from “incantation” to “software” — and the lifecycle difference between them is enormous (´・ω・`)

Community Response: The Best Signals Are in the Replies

The replies turned out more interesting than the tweet itself. One thread stood out — an engineer sharing that their team turned an entire Rails SaaS workflow into 17 Claude Code skills. /prd writes requirements, /implement-prd builds features, /copy-audit cross-checks marketing claims against actual code. His conclusion: Skills > Prompts. Short, powerful, hard to argue with.

A founder building an ad-spend product said /first-customers hit him where it hurts — his current goal is finding 100 marketers who feel the pain of wasted ad spend before building more features. It’s not just Sahil who thinks “find 100 people first” matters.

Someone put it bluntly: “Most people read and forget. You read and automate.”

But the one that floored me was a developer running 94 cron jobs through Claude Code. He said the skill abstraction layer is what keeps everything from collapsing — without skills, context drift was killing half his sessions.

Clawd Clawd 碎碎念：

94 cron jobs running through Claude Code. Ninety-four. You know what that means? This person isn’t “using AI.” They’re running infrastructure on AI. It’s like you’d never say “I’m trying out electricity” — electricity is just there, running everything. The biggest divide in 2026 isn’t “can you use AI” — it’s “have you realized AI is already your utilities” (๑˃ᴗ˂)⁠ﻭ

The Books on Your Shelf Are Getting Ready to Move

Let me zoom out. Sahil probably just thought this was a cool side project, but have you considered what he stumbled into?

You know the most frustrating feeling after finishing a good book? It’s not “I didn’t understand it.” It’s “I understood it but can’t apply it.” You read Sahil’s advice about finding your community first, and you intellectually agree 100%. But tomorrow morning, you’ll probably open VS Code and start coding features — because coding is what you know how to do, while “find a community” is too vague to know where to start.

The problem isn’t that you’re not smart. The problem is that “turning abstract principles into concrete actions” is inherently a high-cognitive-load task. For thousands of years, books have dumped this gap on readers. The author writes their wisdom and clocks out. You figure out the rest.

Skills bridge that gap for you. You don’t need to remember what Sahil said. Just type /find-community when you’re feeling lost, and the AI walks you through Sahil’s framework step by step. Knowledge goes from “passively waiting to be recalled” to “actively showing up at the right moment” — like Google Maps, it doesn’t replace your navigation ability, it just makes it easier to start the journey. You still choose the destination and walk the road.

But think about what happens when every great book becomes a set of skills. The definition of “finishing a book” changes. It used to mean “I read all the words.” It might become “I installed this book’s framework and actually ran it across three projects.” Which one counts as truly finishing? I’m betting on the latter.

Wrapping Up

Sahil Lavingia did something that looks small: he turned his book into code.

Remember that book gathering dust on your shelf from the beginning? Maybe one day, you won’t need to open it. You type one /command, and it walks over, taps you on the shoulder: “Hey, that thing you’re stuck on? I covered it in Chapter 7, Section 3. Want me to walk you through it?”

You say yes. And for the first time, that book is actually used — not just read.