We watched 5+ hours of Claude Code tutorials so you don't have to. Here's everything you need to know to go from zero to building real, functional projects, with timestamped links so you can follow along.
You've probably heard people say Claude Code is the best way to build software with AI right now. And honestly? They're right (especially if you've never coded before and don't know what you're doing). But learning it from scratch can feel like drinking from a firehose.
So we did the hard part for you.
We sat through two of the best Claude Code tutorials on YouTube (over five hours combined), took notes like obsessive grad students, and boiled it all down into one guide. We pulled the clearest explanations and most unique insights from each, and hyperlinked every section to the exact timecode so you can watch along at your own pace.
Your two professors for today:
Both are excellent. Nick goes wider and more technical. Sabrina goes deeper on one practical, end-to-end build. Together, they cover everything.
Both. But if you're short on time:
Nick Saraev's full course | Sabrina Ramonov's full course
Now let's get into it.
Before we get our hands dirty, let's make sure we're on the same page.
Claude Code is Anthropic's command-line tool that lets Claude (the AI) write, edit, and run code directly on your computer. Instead of copying and pasting code from a chatbot into your files, Claude Code works inside your project, reading your files, making changes, and even running tests.
Think of it like this: regular Claude is a consultant you email back and forth. Claude Code is a consultant who sits at your desk and does the work while you watch.
As Nick explains in his introduction, Claude Code runs in the terminal (that black screen with text that developers use). But don't let that scare you; both tutorials show you exactly how to set it up without needing a computer science degree.
Sabrina's course has the most beginner-friendly installation walkthrough. She walks through every click, starting from a fresh VS Code setup, which is nice if you've never opened a code editor before.
Here's what you need:
Once those are in place, you install Claude Code with a single command: npm install -g @anthropic-ai/claude-code. Then type claude in your terminal and you're off to the races.
"IDE" stands for Integrated Development Environment. Fancy name for "the app where you write code."
Nick dedicates a whole section to comparing IDEs. He covers VS Code (free, popular, tons of extensions) and Antigravity, a newer editor built specifically for AI-assisted coding. His recommendation: VS Code for beginners, Antigravity if you want something purpose-built for Claude Code.
Sabrina keeps it simple and sticks with VS Code the entire time. For most people reading this, that's the right call.
One thing Sabrina emphasizes that many tutorials gloss over: permissions and safe mode.
When Claude Code runs on your computer, it can read files, write files, and execute commands. That's powerful, but it means you should understand what you're allowing. Sabrina walks through the permission prompts one by one, explaining which ones to approve and when to be cautious.
Key takeaway: Start in safe mode, where Claude asks before running any command. You can loosen the leash later once you're comfortable.
Before you start building, you need to understand Claude Code's two brain modes. Sabrina covers this clearly in her Plan vs. Edit breakdown:
Plan Mode is like asking Claude to think out loud. It analyzes your project, proposes a plan, and waits for your approval before touching anything. Great for complex tasks where you want to review the approach first.
Edit Mode is the default. Claude reads your request, writes the code, and makes changes directly. Faster, but you're trusting it more.
Pro tip from Sabrina: Start with Plan Mode for anything complex. Switch to Edit Mode for small changes and quick fixes. You can toggle between them mid-conversation.
Nick adds another layer in his advanced section: you can also bypass permission prompts entirely with --dangerously-skip-permissions for tasks you fully trust. Useful for automation. Obviously, use with care.
Ever had a conversation with ChatGPT where it suddenly forgot what you were talking about? Same thing happens with Claude Code, and understanding why is the difference between a frustrating experience and a productive one.
Sabrina's context window section explains the concept simply: Claude can only "see" a limited amount of text at once (its context window). Bigger projects = more files = more text competing for that limited space.
Nick goes even deeper in his context management chapter. His strategies for staying within limits:
/compact command to summarize the conversation and free up spaceNick also covers token management strategies in detail. He shows how to monitor and reduce your token usage (i.e. how much you're spending per session). Worth watching if you plan to use Claude Code regularly.
Both instructors agree: the CLAUDE.md file is the single most important thing in your Claude Code setup. It's basically a cheat sheet that Claude reads every time it starts working on your project.
Nick's CLAUDE.md walkthrough explains the concept as your "project brain." You write instructions like: "This is a React app. We use TypeScript. Our API endpoints live in /api. Never delete test files." Claude reads this before every task, so it always has context about your project.
Sabrina shows the practical side in her CLAUDE.md documentation section near the end of her course. She walks through how she documented her AI Marketing Officer project so anyone (or any future Claude session) could pick up where she left off.
What to put in your CLAUDE.md:
npm test")Think of it like onboarding instructions for a new hire who's extremely smart but has zero context about your company. That's Claude.
Here's where the two courses diverge in the best possible way.
Nick's first project build is a great demo of Claude Code's raw speed. He asks Claude to build a complete web application from a simple description, and within about 15 minutes, he has a working app.
He follows this up with a deeper dive into website design approaches using Claude Code, showing how to iterate on designs by giving Claude feedback like you'd give a human designer. ("Make the header bigger." "Move the CTA above the fold." "This looks like it was made in 2005, fix it.")
Key lesson from Nick: You don't need to describe every pixel. Give Claude a clear vision of what you want, let it build a first draft, then iterate. The back-and-forth is where the magic happens.
He also emphasizes verification, reminding viewers to always check Claude's output. AI is fast, but it can also be confidently wrong. Trust, but verify.
Sabrina takes a completely different path that's arguably more relevant for non-developers. Her entire course is one end-to-end build: an AI Marketing Officer that can research topics, draft social media content, add visuals, match your brand voice, and publish across platforms.
Type one sentence, and the AI researches the topic, writes a week's worth of content, and schedules it. That's the pitch, and she actually builds it.
Her first skill build (the "Post" skill) teaches Claude how to create a social media post from a YouTube video. She walks through connecting the Blotato API for generating images, testing the skill live, and debugging API errors in real-time.
That debugging section is gold, by the way. Most tutorials only show you the happy path. Sabrina shows you what happens when things break (because they will) and how to fix them.
"Skills" are reusable instructions you save so Claude can perform specific tasks without you re-explaining them every time.
Nick covers the theory and structure of skills thoroughly. He explains how skills live in your .claude directory (your project's hidden config folder) and walks through building a new skill from scratch, specifically a lead scraping tool for business development.
Sabrina shows the same concept but applied to content creation. Her Post skill teaches Claude how to take a YouTube video URL, extract key insights, write a social media post, generate a matching image via the Blotato API, and format everything for publishing.
The pattern is the same:
Once a skill is built, you can use it over and over. Nick's lead scraper can process hundreds of leads. Sabrina's Post skill can crank out content from any YouTube video. You build it once and reuse it forever.
This is where Sabrina's course really shines. In her brand voice section, she shows how to teach Claude to write in specific styles by giving it examples.
She demonstrates two voices:
The technique is straightforward: feed Claude examples of writing you like, tell it to analyze the patterns (sentence length, vocabulary, structure, tone), and save those patterns as part of your skill. Now every piece of content Claude writes sounds like you, not like a robot.
If you're using Claude for writing and the output sounds generic, this section alone is worth the watch.
Here's something clever. Sabrina's quality gates section shows how to set up "hooks," which are automatic checks that run before or after Claude performs an action.
Think of them like spell-check, but customizable. You can create hooks that:
This ties into Nick's coverage of hooks and slash commands in his advanced section, where he explains the broader hook system. But Sabrina's practical example of using hooks specifically for content quality is one of the most immediately useful things in either course.
This is where things get wild. Both courses cover sub-agents, which let Claude spin up additional instances of itself to work in parallel.
Sabrina's sub-agent tutorial is the easiest to follow. She uses sub-agents to publish content to Twitter and LinkedIn simultaneously. Instead of Claude writing a post, publishing to Twitter, waiting, then publishing to LinkedIn... it spawns two sub-agents that handle both platforms at the same time.
Nick takes sub-agents further. His introduction to sub-agents explains how to turn any skill into a sub-agent, effectively giving Claude a team of specialists it can delegate to.
He demonstrates this with an email classification system that can process and sort hundreds of emails by spawning multiple sub-agents to handle them simultaneously. Then he goes even further with agent teams, where multiple sub-agents coordinate on complex, multi-step projects.
The simple version: Sub-agents let you parallelize work. Instead of one Claude doing 10 tasks sequentially (slow), you get 10 Claudes each doing one task at the same time (fast).
Sabrina's second big build is the Plan Week skill, which automates an entire week's content calendar.
Here's the workflow she builds:
The full weekly content plan review shows the final output: a complete week of social media content from one input. It's impressive, and she built it all live on camera.
She also generates content from Perplexity search results, not just YouTube videos. So you can feed it trending topics and get posts without needing source videos.
Nick's course covers several advanced topics that Sabrina doesn't touch. That makes them perfect complements.
Nick's MCP section explains how to connect Claude Code to external tools and services. MCPs are like plugins that give Claude superpowers; access to your Gmail, Google Drive, databases, APIs, and basically anything else.
He walks through the full MCP setup, including practical warnings about token usage when using MCPs (spoiler: they can get expensive if you're not careful).
Nick's GitHub section shows how Claude Code works with version control (i.e. the system developers use to track changes and collaborate). If you're working on a team or want to keep a history of changes, this is essential.
One of the most powerful (and underrated) features Nick covers is Git worktrees. In plain English: worktrees let you run multiple Claude Code instances on different branches of the same project simultaneously, without them stepping on each other's toes.
Imagine asking one Claude to build Feature A while another builds Feature B and a third fixes bugs. All at the same time, all on the same project, no conflicts. That's worktrees.
Nick wraps up with deploying projects to the cloud using Modal. This takes your project from "works on my computer" to "live on the internet." If you're building something you want other people to use (or something you want to sell), this is the final step.
After 5+ hours of tutorials, here's what stuck with us:
From Nick:
/compact aggressively to manage your context window. (Watch)From Sabrina:
Grant Harvey is the Lead Writer of The Neuron, where he continues to lead the publication's daily coverage of AI news, tools, and trends.
Don't fall behind on AI. Get the AI trends & tools you need to know. Join 700,000+ professionals from top companies like Microsoft, Apple, Salesforce and more.
Property of TechnologyAdvice. © 2026 TechnologyAdvice. All Rights Reserved
Advertiser Disclosure: Some of the products that appear on this site are from companies from which TechnologyAdvice receives compensation. This compensation may impact how and where products appear on this site including, for example, the order in which they appear. TechnologyAdvice does not include all companies or all types of products available in the marketplace.