You will be able to
- Install and configure Claude Code in a terminal and run it alongside your project
- Write clear, specific prompts that give a coding agent the context it needs
- Create a CLAUDE.md project memory file and keep it lean
- Use plan mode to break a feature into reviewable steps and verify the result
- Manage the context window with clear and compact, and control usage and cost
- Extend Claude Code with MCP servers and decide which ones are worth their context cost
Good to know first
- At least three months of experience building React applications
- Comfortable with JavaScript and TypeScript, including components, state, props, and hooks
- Basic terminal use and running npm commands
- Basic Git usage, such as adding files and making commits
Syllabus
10 lessons and a final project
- 1
Course overview, repositories, and how to learn from this course Free
This lesson shows what you will build in the Claude Code course: a full stack AI-powered support ticket system taken from idea to live deployment. It also explains the course structure, the two repositories you will clone, and the hands-on way to learn that makes the course work.
4 build steps2 exercises4 questions - 2
What Claude Code is and why it changes development Beyond Pass
You learn what makes Claude Code an agentic coding tool, how it differs from a chat assistant that only suggests code, and why software engineering still matters. You also see the carpenter analogy that explains where coding agents fit in your work.
2 exercises4 questions - 3
Installing Claude Code and setting up the terminal Beyond Pass
You will learn how to get a Claude plan or API access, install Claude Code from the terminal, and clone a starter project. You will also set up a comfortable terminal layout in VS Code and use a few shortcuts (!, /, @, /terminal-setup).
7 build steps3 exercises4 questions - 4
Running the expense tracker and granting tool permissions Beyond Pass
You learn how to start the expense tracker by asking Claude Code to run it, and how Claude asks your permission before using a tool such as the bash shell. You also learn how to read the command it wants to run and choose the right permission option.
1 code6 build steps2 exercises3 questions - 5
Creating CLAUDE.md project memory and committing with Claude Beyond Pass
You will learn how the /init command creates a CLAUDE.md file that gives Claude Code lasting memory of your project's commands, architecture and conventions. You will also see how to review the file in source control and ask Claude Code to write a clean commit.
1 code5 build steps2 exercises4 questions - 6
Writing Effective Prompts to Find and Fix a Bug Beyond Pass
You will learn three rules for writing good prompts to a coding agent: be clear and specific, give context, and be concise. You will also see how a follow-up prompt can turn a quick patch into a real fix of the root cause.
6 build steps1 exercise4 questions - 7
Refactoring components and explaining unfamiliar code with Claude Code Beyond Pass
You will learn how to split a growing React component into smaller, focused child components by asking Claude Code to do it, and how to keep the project notes (CLAUDE.md) up to date. You will also learn how to ask Claude to explain an unfamiliar line of code you have selected.
2 code6 build steps3 exercises3 questions - 8
Planning and building a delete feature in plan mode Beyond Pass
You will see how to use Claude Code's plan mode to design a small feature before any code changes, then review the plan, approve it, and check the result. You will also learn why small, reviewable feature steps keep AI-written code consistent.
2 code6 build steps3 exercises3 questions - 9
Managing the context window with clear and compact Beyond Pass
You will learn what Claude Code's context window is and how to see how full it is with /context. You will also learn when to use /clear for unrelated tasks and /compact to summarize related work.
4 build steps3 exercises3 questions - 10
Tracking usage and costs, and extending Claude with MCP Beyond Pass
You will learn how to check your Claude Code usage, why a longer context window costs more, and how the Model Context Protocol (MCP) lets Claude use outside services like GitHub through MCP clients and servers. You will also learn why adding every MCP server you can find is a bad idea.
3 build steps1 exercise4 questions
Final project
Ship a feature with a lean Claude Code workflow
Take the expense tracker starter and add a new feature, such as a category filter or a monthly summary. Use Claude Code the way the course teaches: project memory, specific prompts, plan mode, and context control. You will end with a reviewed commit and a short log of what each technique did for you.
Its 7 tasks open with Beyond Pass.
Free preview · Lesson 1 in full
Course overview, repositories, and how to learn from this course
This lesson shows what you will build in the Claude Code course: a full stack AI-powered support ticket system taken from idea to live deployment. It also explains the course structure, the two repositories you will clone, and the hands-on way to learn that makes the course work.
Draw and write next to the video.
The idea
Claude Code is an AI coding agent: you describe what you want in plain language, and it reads your project, writes and changes code, and runs commands with you. The course teaches you to use it to build real production software, not toy demos. The video says engineers at Anthropic already use it to write up to 90% of their code, and the goal here is to learn the engineering habits that make AI-written code trustworthy.
The main project is a full stack support ticket system called the help desk. Customers send in tickets, agents manage them with filtering, sorting and pagination, and users log in with role-based access control so the right people see the right things. AI features then make the system smarter: summarizing long conversations, polishing agent replies to sound professional and on brand, classifying new tickets, and automatically answering questions from a knowledge base. Incoming emails are picked up by a background AI agent that can reply within seconds, and anything it cannot resolve is escalated to a human agent.
The course has nine sections. Section 2 teaches Claude Code on a small expense tracker that has a bug and messy code. Sections 3 to 8 build the help desk: planning and setup, authentication, user management, ticket management, AI features, and email integration. Section 9 covers production: error logging, UI polish, Docker, and deployment. The finale connects Claude Code to GitHub Actions, so an issue you open can be fixed by Claude, reviewed by you, merged, and deployed.
You need about three months of React experience (JavaScript, TypeScript, components, state, props, hooks, and calling APIs). The back end uses Express rather than Next.js, so front end and back end are clearly separate, as in many real projects. Knowing Express helps but is not required, and the concepts apply to Next.js too. The focus is the engineering mindset, not the tool, so the same ideas apply to other coding agents.
There are two repositories. The expense tracker is the playground for section 2, so clone it first. The help desk is the full project built through the rest of the course. The author commits after each lesson, so you can compare your code with the exact code produced at each step.
This is a software engineering course, not a demo you watch. Keep your editor and terminal open and build along. Expect to read every line the AI writes, spot issues, and refactor. The pace speeds up later, so doing it yourself is the only way to keep up. Take the course from start to finish, because each lesson builds on the one before.
Key points
- Claude Code lets you build a real full stack app with AI, but you still review every line of generated code. 0:00
- The project is a support ticket system with authentication, role-based access control, and AI features. 0:21
- The course is in nine sections: foundations, planning, auth, users, tickets, AI features, email, production. 0:41
- The back end uses Express so front end and back end stay clearly separated; the concepts apply to Next.js too. 2:37
- Clone two repositories: the expense tracker for the foundations, and the help desk for the main project. 4:48
- Build along with the course: have your editor and terminal open, and do the work yourself. 7:32
Build along in the practice terminal
- Check your background: you should be comfortable building React apps (components, state, props, hooks) and calling APIs. If not, strengthen React first. 2:37
- Clone the expense tracker repository now. You will use it in section 2 as your playground for learning Claude Code. The link is in the video description. 5:48
- Clone the help desk repository too. You will build it step by step in the rest of the course. The link is in the video description. 6:08
- Keep an editor and a terminal open while you watch, and pause to try each step yourself. 7:32
Practice
In your own words, explain why the course says you should review every line of AI-generated code instead of accepting it as it is.
Solution
- The AI can produce code that works in a demo but has bugs, poor structure, or missing edge cases.
- A real engineer checks the code, finds issues, refactors it, and makes sure it follows solid practices.
- This is what makes the app safe to ship to real customers.
Model answer: AI code can look right but still contain bugs, messy structure, or gaps. Reviewing each line lets you find issues, refactor, and follow sound engineering practices, so the app is safe to ship to real customers.
Set up your learning environment before section 2: create a folder for the course, clone the expense tracker repository into it, and open it in your editor. Then list the files in the folder to confirm the clone worked.
Solution
- Create a folder, for example programming-course, and move into it in your terminal.
- Run git clone with the expense tracker URL from the video description.
- Move into the new folder and list its files with ls (macOS or Linux) or dir (Windows).
- Open the folder in your editor and check that a project with source files is there.
The expense tracker project folder is cloned and open in your editor, with its source files listed.
On your scratchpad
Draw a simple map of the course: a line of nine boxes labelled with the section names, with the expense tracker box marked as the warm-up and the help desk as the main project. Write one sentence under it describing what you will build.
Quiz
What is the main project you build through the course?
- A calculator app that runs in the terminal
- A full stack AI-powered support ticket system
- A static portfolio website
- A Next.js blog with no back end
Show the answer
B. A full stack AI-powered support ticket system · The video says the course builds a full stack AI powered support ticket system, the kind you would ship to real customers.
0:15 watch the momentWhy does the course use Express for the back end instead of Next.js?
- Express is faster than any other framework
- It keeps front end and back end clearly separated, as many real projects are
- Next.js cannot connect to a database
- The course forbids Next.js
Show the answer
B. It keeps front end and back end clearly separated, as many real projects are · The author says many developers prefer complete separation of front end and back end, and Express makes each side visible without framework magic.
2:37 watch the momentWhich repository should you clone first, and why?
- The help desk, because it is the final project
- The expense tracker, because it is the playground for section 2
- Neither, you only need the course slides
- Both at the same time with no order
Show the answer
B. The expense tracker, because it is the playground for section 2 · The expense tracker is used in the next section to learn the foundations of Claude Code, so it is cloned now.
5:48 watch the momentHow does the author suggest you get the most from the course?
- Watch it passively and let AI build the app
- Build along, review the AI code, and do the work yourself
- Skip to the AI features section
- Only read the code on GitHub
Show the answer
B. Build along, review the AI code, and do the work yourself · The author says the course is not passive: you build along with your editor and terminal open and go through the code yourself.
7:32 watch the moment
Lessons 2–10 and the final project
They open with Beyond Pass: every lesson of every course, in full. One payment: $14.99 for 90 days or $29 for a year. No subscription.
Made by Beyond from the video “Claude Code Tutorial - Build Apps 10x Faster with AI” by Programming with Mosh. The video plays in YouTube's own player. Watch it on YouTube · All lessons