The Claude Code Masterclass.

Build and ship your first real project, in Claude Code or Codex.

Steps

1. Install
2. Prompt
3. Skills
4. Deploy
5. Database
6. Automate

Tools

Claude Code, Codex, GitHub, Vercel, Supabase

• 0.1Claude Code Foundations

  • Set up Claude Code, prompt well and install skills
  • Design a site, push it to GitHub and go live on Vercel
  • Leads in Supabase, email alerts, a private CRM and a research automation

• 0.2Codex Foundations

  • Context, skills, plugins and planning in Codex
  • Front end, APIs, AGENTS.md and a database
  • Email, testing, deployment, and Codex with Claude Code

The working knowledge every Claude Code user needs, done properly.

Steps

1. Context
2. Skills
3. Instructions
4. Models
5. Tools
6. Git
7. Review

Tools

Claude Code, Codex, GitHub

• 1.1Tokens, Context Window & Caching

  • What actually fills the context window
  • /compact vs /clear vs a new chat
  • Spot context rot before it costs you

• 1.2Skills

  • How skills work, and when they actually trigger
  • What happens when similar skills collide
  • Build and test one with the skill creator

• 1.3CLAUDE.md & AGENTS.md

  • What belongs in project instructions, and what doesn’t
  • Why less is more with current models
  • My production CLAUDE.md, walked through

• 1.4Models & Effort Levels

  • Which model for which job, Claude and Codex
  • When more effort is worth it
  • Read your usage and cut your bill

• 1.5Ultracode

  • What Ultracode and dynamic workflows actually do
  • When a job is big enough to need it
  • Set agent count, model and budget

• 1.6APIs, CLIs, MCPs & Plugins

  • What each one is, and when to reach for which
  • Plugins: skills, hooks and MCPs in one install
  • Install from a marketplace, and vet it first

• 1.7Hooks & Mods

  • Hooks: automatic checks on every edit
  • Mods: change how Claude Code looks and behaves
  • Mods aren’t sandboxed, so read before you install

• 1.8Git & GitHub

  • Commit, branch and roll back when Claude breaks something
  • Worktrees for trying two ideas at once
  • Back up to GitHub, with secrets left out

• 1.9Claude Projects

  • Hand Claude a stream of tasks to run as parallel threads
  • Shared instructions and memory for every thread
  • When a Project beats a single session

• 1.10Claude Remote

  • Run Claude Code from your phone
  • Check on long jobs from anywhere
  • Remote Control vs Dispatch vs Channels

• 1.11Code Review & Multi-AI Setups

  • Why the model that wrote the code shouldn’t grade it
  • The claudex loop: Claude builds, Codex reviews until it’s approved
  • When a second AI is worth it, and when it isn’t

• 1.12Capstone projectBuild and share your own skill or mod

  One skill or mod that fixes something you do every week, on GitHub so others can install it.

Websites that look designed, not generated.

Steps

1. Inspiration
2. Prototype
3. Tweak
4. Motion
5. 3D

Tools

Impeccable, FFmpeg, Blender

• 2.1The 3-Step Web Design Workflow

  • Inspiration: build your own design library
  • Prototype: three directions that really differ
  • Tweak: fine-tune the winner with a tweak panel

• 2.2Web Design Skills

  • Impeccable and other design skills
  • Component libraries and copy-prompt sites
  • Layout, type and color systems Claude can follow

• 2.3Scrolling Animations & Motion Design

  • Scroll-scrubbed video, cut into frames with FFmpeg
  • Fly through a generated world as you scroll
  • Load-in, reveals and parallax, checked on mobile

• 2.4Blender & 3D Web Design

  • Model a hero asset in Blender
  • Export it light enough for the web
  • Add interaction without wrecking load time

• 2.5Capstone projectYour design library and website

  Fill a design library with inspiration you like, then build and launch your site from it.

Give Claude a memory it can search, from your notes to production RAG.

Steps

1. Memory
2. Obsidian
3. RAG
4. Production
5. GitHub

Tools

Obsidian, LightRAG, Graphify, Railway, GitHub

• 3.1Claude, Memory & Second Brains

  • Where memory lives: CLAUDE.md, built-in memory and files
  • Build a second brain: a vault, a brain dump, an interview
  • Point every session at your vault so it can search it

• 3.2Obsidian & Karpathy “RAG”

  • Karpathy’s structure: raw, wiki and outputs
  • Claude reads the map first, then the right file
  • Capture sources with the Obsidian Web Clipper

• 3.3RAG Basics

  • How retrieval works, and when plain files are enough
  • LightRAG and RAG-Anything
  • Knowledge graphs with Graphify

• 3.4Production-Level RAG

  • Take a local RAG system to the cloud
  • Security and sign-in for your team
  • Deploy on Railway, with documents added by upload or Drive

• 3.5GitHub as Memory

  • Version your knowledge with GitHub
  • Review what changed and why
  • Private vs shared knowledge

• 3.6Capstone projectMap your second brain

  Your real vault, every folder explained, and Claude answering a real question from it.

Claude working on a schedule, in the browser and on your desktop.

Steps

1. Schedule
2. Browser
3. Use cases

Tools

Claude in Chrome, Playwright

• 4.1Automations & Routines

  • Schedule Claude in the cloud, the desktop app or with /loop
  • Triggers: a time, an API call or a GitHub event
  • A test run and drafts before it runs alone

• 4.2Browser Automations & Computer Use

  • The built-in browser, Claude in Chrome, Playwright and computer use
  • Playwright for repeat runs, computer use for desktop apps
  • Approving sites and apps, and watching for prompt injection

• 4.3Automation Use Cases

  • My automations: morning intel, lead research, content cascade
  • The five parts: trigger, gather, do it, save it, review
  • Is it worth automating? Four quick checks

• 4.4Capstone projectYour daily automation

  Found from your own work, then tested, scheduled and shared on GitHub.

Videos, copy and carousels, made with Claude.

Steps

1. Motion
2. Edit
3. Write
4. Images
5. Carousels

Tools

HyperFrames, Higgsfield, Nano Banana, Seedance, Kling

• 5.1Motion Graphics & JavaScript

  • Claude writes the animation, a browser renders every frame
  • Four levels: prompt, review loop, references, /animate
  • No AI video model, so it only costs your Claude usage

• 5.2Video Editing

  • How HyperFrames turns your clips into an edit
  • The /claude-edit flow: cut, transcript, beat plan, build
  • What to film, and installing claude-edit

• 5.3Claude & Copywriting

  • What AI writing sounds like, and how Humanizer catches it
  • Your own voice skill, tested before and after
  • One source rewritten for LinkedIn, X and email

• 5.4AI Images & Video

  • Where the models live, and what they cost
  • Consistent characters with a character sheet and references
  • Let Claude write the prompt, then change one thing at a time

• 5.5Carousels for Social Media

  • The carousel engine we post from, free on GitHub
  • Ranked hooks and a brief before any image exists
  • Real screenshots and exact text pasted in, never drawn

• 5.6Content & Marketing Skills

  • The skills we use every week, and what each one does
  • Install them from one repo, then build your own

• 5.7Capstone projectMake a full video with Claude

  An explainer written in code, your own video edited by Claude, or both on a split screen.

Real apps with accounts, data and a design you can explain.

Steps

1. Plan
2. Clone
3. Build
4. Data

Tools

Supabase, SQLite, Postgres

• 6.1System Design

  • Follow one request through a real app
  • The outside services and AI models it calls
  • Sign-in, and what runs when nobody clicks

• 6.2Clone Any App

  • Research a real product’s feature set
  • Set clear success criteria with /goal
  • Build it locally, then make it yours

• 6.3App Building

  • The architecture of my AI Twitter engagement engine
  • Request-response apps vs background jobs
  • When you need a real multi-user app

• 6.4Databases

  • SQLite, Postgres and vector search, and when to use each
  • Supabase, run by Claude Code through its CLI
  • Row-level security on every table, tested as two users

• 6.5Capstone projectClone it, make it yours

  Clone an app you would use, add your own feature and a database, and break down its system design.

Your skills, memory and automations working as one system.

Steps

1. OS
2. Loops
3. Graphs
4. Share

Tools

Jev, Obsidian, GitHub

• 7.1Agentic OS

  • The architecture, with Jev at the core
  • Skills and Obsidian memory
  • A dashboard on top, and how to install it

• 7.2Loop Engineering

  • What a loop is, and when not to use one
  • /goal, clear goals and a separate verifier
  • Turn your loop into a skill

• 7.3Graph Engineering

  • Trigger, task and success check: the parts of every loop
  • One agent per task, side by side, then a review
  • When a graph beats a loop, and when it doesn’t

• 7.4Sell / Share Your AIOS

  • A clean copy of your OS, without keys or client data
  • Shape the repo like our starter: one install prompt
  • Walk someone through installing it

• 7.5Capstone projectAgentic OS showcase

  Share your custom Agentic OS setup.

Turn what you’ve built into an AI agency.

Steps

1. Offer
2. Story
3. Network
4. Content
5. Call
6. Audit
7. Proposal

Tools

Stripe, Calendly, LeadShark, ManyChat

• 8.1Building Your AI Agency

  • Fundamentals beat the latest tool
  • Why consistency matters more than tech skills
  • Your 90 days, step by step

• Companion courseThe 90-day AI Agency Blueprint30+ lessons

  Inside Chase AI+: build your AI business and land clients. It covers:

  • The 90-day roadmap
  • AI solutions that sell
  • Your business stack
  • Your story & network
  • Social media & LinkedIn
  • The discovery call
  • The AI audit
  • Pricing, proposals & closing
  • Kickoff to hand-off

Everything you need to keep building.

• 9.1Master Skill List

  • Every skill in the course, in one list
  • Where each one lives, and how to install it