Orientation

How to use this program

This Bible is a complete course, not a reference doc to cherry-pick. The modules go from the simplest (talking to Claude in a browser) to the most advanced (orchestrating dozens of agents, building a second brain, composing multi-vendor system prompts).

Three ways to use it:

• Linear: follow the modules in order. Each module assumes the previous one is learned.
• By level: filter by Beginner / Intermediate / Advanced / Expert in the top bar.
• Targeted: use search to jump straight to a topic (hooks, MCP, KV-cache, prefill...).

Every lesson has the same anatomy: an explanation, key takeaways to remember, often a demo you can copy, and a one-question quiz to check yourself. Your progress is saved in the browser (tick "Read" at the bottom of each lesson).

The through-line: you don't learn a list of buttons, you learn to think with a model. The same reasoning serves you in the chat, in Cowork, in Claude Code, and with any other LLM.

The Claude landscape on one map

Anthropic is three surfaces for the same brain (the Claude model):

• Claude.ai (chat): the browser or the app. Conversation, Projects, artifacts. Everyone's entry point.
• Claude Cowork: the agentic mode in the desktop app. Claude reads, edits, creates and organizes files on your machine, plans and executes multi-step tasks autonomously. Went GA in April 2026.
• Claude Code: the command-line agent for code. CLI, IDE, web. It is the tool writing this very document.

Underneath all three: the same Claude API (formerly the Anthropic API) and the same model family (Opus, Sonnet, Haiku). Learning to talk to one is learning to talk to all. What changes is the tools the model has on hand (read a file, run a command, search the web) and the context you give it.

The mental rule to keep everywhere: an LLM is a function that takes text (the context) and returns text (the answer). The whole craft is controlling what goes into the context.

Your learning path through this Bible

This course is organized into modules (topic clusters) and lessons inside each module. Every lesson carries a level tag: Intro, Intermediate, or Expert. You can read them in order or jump to whichever level matches your current comfort.

Each lesson follows the same four-step loop so you always know what to expect:

1. Read: the core explanation, plain and concise.
2. Demo: a real prompt, command, or comparison you can copy and try.
3. Exercise: a short hands-on task, doable in a few minutes.
4. Quiz: one question to lock in the concept before moving on.

Progress is tracked locally in your browser (no account needed). A small indicator marks each lesson as visited or completed. You can reset tracking at any time from the settings panel. Because everything runs as a single HTML file, your progress is tied to that file on your device.

How each lesson is built

Every lesson in this course follows the same five-part structure so you always know what to expect and where to find things.

The five parts are:

1. Explanation: the core concept, written in plain language with jargon defined on first use.
2. Key takeaways: three or four bullet points that capture the essentials you should remember.
3. Demo: a real, copyable example, such as an actual prompt you can paste into Claude.ai, a real claude command you can run in your terminal, or a side-by-side comparison table.
4. Exercise: a short hands-on task you can complete in a few minutes, with a worked answer you can reveal once you have tried.
5. Quiz: one multiple-choice question that checks whether the main idea landed.

Knowing the structure lets you skim or deep-read depending on how much time you have. If you already understand the explanation, jump straight to the demo or the exercise.

Set up your practice space

The fastest way to learn Claude is to interact with it immediately. You need three things: a Claude chat tab, a plain text file to take notes, and optionally the Claude Code CLI (command-line interface, meaning a tool you run in a terminal window) for hands-on coding practice.

Start by opening Claude.ai in your browser. This is the chat interface, and it needs no installation. Create a free account if you do not have one yet. Keep this tab open throughout every lesson so you can test prompts the moment you read about them.

Your second tool is a scratch notes file: a plain .txt or .md file on your desktop where you paste prompts that worked well, outputs that surprised you, and questions to explore later. Reviewing this file once a week is worth more than rereading any lesson.

If you want to go further, you can install Claude Code, the terminal-based coding agent. It is optional for this course but unlocks hands-on coding exercises. The install is one command:

• Make sure Node.js (version 18 or later) is installed on your machine.
• Run the install command shown below in your terminal.
• Authenticate with your Anthropic account when prompted.
• Type claude --version to confirm it is working.