Advanced: Claude Code and Agentic AI
Early access, Chapter 1
This tier is still growing. You get full access to every lesson published today, plus every new lesson as we release it, at no extra cost. Coming next: Gmail, Calendar and GitHub MCP servers; building your own MCP; real-world Claude Code case studies.
The frontier. Claude Code, plan mode, slash commands, CLAUDE.md, MCP servers, and multi-agent workflows. Still expanding as the tools evolve.
Meeting Claude Code
What Claude Code actually is (and isn't)
8 min · Claude Code is not a chatbot. It is an AI that lives in your terminal, reads your files, and does real work. Here is the plain-English version.
Installing Claude Code on macOS and Windows
10 min · A step-by-step install guide for non-developers. Covers the terminal, Node.js, your API key, and the common gotchas.
Your first three commands
10 min · Three real tasks to try in your first ten minutes with Claude Code. You will finish with a working example, not just theory.
The safety model, and why it asks permission
8 min · Claude Code lets you go faster once you know when to trust it. Here is the mental model and a practical settings recipe.
Working effectively with Claude Code
Plan mode, and when to use it
8 min · Plan mode turns Claude Code from "ready to act" into "ready to think". The single biggest productivity win once you are past day one.
The slash commands you will actually use
10 min · Claude Code has a lot of slash commands. You only need eight of them for everyday work. Here they are, in a table.
CLAUDE.md, teaching Claude about your project
8 min · A small Markdown file in your project that Claude Code reads on every session. The single cheapest upgrade to every prompt you will ever write.
Tokens, context, and not burning through your plan
9 min · A practical guide to how Claude Code uses its context window, why you hit limits, and what to do about it.
Writing prompts that get Claude Code to do the right thing
10 min · Claude Code prompts are different from chatbot prompts. Here is what changes, and how to write prompts that land first time.
MCP servers
What is MCP? The universal connector, explained
8 min · The one protocol that lets any AI talk to any tool. MCP in plain English, with real examples.
Adding your first MCP server (Notion)
12 min · A hands-on walkthrough. By the end, Claude Code can search, read, and create pages in your Notion workspace on your behalf.
Playwright MCP: give Claude a browser
14 min · Set up Playwright so Claude can navigate real websites, fill in forms, screenshot pages and pull data. The most visible wow factor of the MCP world.
Gmail MCP: let Claude read and draft your email
12 min · Connect Gmail to Claude Code. How to authenticate safely, start with drafts only, and graduate to real inbox workflows once you trust it.
The MCP catalog: 8 more servers worth knowing
15 min · A reference for eight MCP servers beyond Notion, Playwright and Gmail. What each one does best, when it is worth installing, and when it is overkill.
Agents and parallel work
What makes a task a good "agent" task
8 min · Not every job is a good fit for autonomous AI. Here are the five signals of tasks where agents save you time, and the five where they will cost you more than they save.
Running agents in the background
8 min · Fire off long jobs and let them run while you do something else. A practical guide to background agents, and how to avoid the common traps.
Using multiple agents, and when it is the right move
9 min · Two or three agents working in parallel can finish a job in a fraction of the time. Or they can tangle each other up. Here is how to tell which you will get.
When NOT to use Claude Code (or any agent)
7 min · The best Claude Code users are the ones who know when to close the terminal. A short, honest guide to the jobs AI should not do for you.