Claude · Learning Map

Claude & AI-Assisted Development

This guide takes you from "what is a language model?" to designing agentic development workflows. It does not just list commands: each page explains why a capability exists, shows how it works, and says where it fails. One Java/Spring Boot project runs through the whole series, and every product claim links to the vendor source it came from.

Parts

Chapters published

Information current as of: 2026-10-30

The Journey

From a Language Model to an Engineering Workflow

The series moves through six steps. Each step reuses the one before it, so you never meet an agent feature before you know what an agent is.

The learning journey

Reading Paths

Pick the Path That Matches You

🌱 Beginner

You write code but have not used AI tools, or only chatted with one.

🔧 Intermediate

You use Copilot or ChatGPT and want to understand agentic tools.

🚀 Advanced

You already use an agent daily and want workflows and governance.

Chapters

The Seven Parts

All fourteen chapters are published. Each page carries its own date, sources and exercises.

Part 1 · Understanding the AI Revolution

Chapter 1 · The World of AI — LLMs, tokens, context windows, hallucination.
Chapter 2 · How AI Changed Software Development — from autocomplete to agents.

Part 2 · Understanding Coding Agents

Chapter 3 · What Exactly Is a Coding Agent? — the agentic loop and its parts.
Chapter 4 · The Coding Agent Ecosystem — tools versus models, eight tools compared fairly.

Part 3 · Understanding Anthropic and Claude

Chapter 5 · Introducing Anthropic and Claude — the model tiers, cost and the three ways in.
Chapter 6 · Claude Code: A Different Way to Develop Software — harness, tools, permissions, Git.

Part 4 · Getting Started

Chapter 7 · Installation and First Steps — install, sign in, create the bookshelf project.
Chapter 8 · Learning to Communicate with Claude — prompts, plans, review, iteration.

Part 5 · Becoming Productive

Chapter 9 · Working with Real Projects — ten scenarios with real outputs.
Chapter 10 · Core Capabilities — CLAUDE.md, plan mode, permissions, skills, hooks, MCP, subagents, worktrees.

Part 6 · Advanced Agentic Development

Chapter 11 · Advanced Workflows — parallel worktrees, verification gates, CI, long-running tasks.
Chapter 12 · Building Your Own Agentic Environment — a hand-written MCP server, plugins, enterprise policy.

Part 7 · Professional Usage

Chapter 13 · Security, Limitations and Responsibility — real defensive tests, and where they fail.
Chapter 14 · The Future of Software Engineering — a labelled view of evidence, not forecasts.

Part 8 · Hands-On: Prompts

Chapter 15 · The Prompt Library — 73 reusable prompts in 13 categories, 29 with real answers and an honest check.
Chapter 16 · Prompt Recipes & Builder — five recipes run for real, and a builder to compose your own.
Chapter 17 · Reading Claude's Output — a real session opened up: tool calls, cost, denials, self-reports.

Part 9 · Hands-On: Skills

Chapter 18 · Your First Skill — three skills built and run for real, plus a troubleshooting experiment.
Chapter 19 · A Skill Pack for a Real Project — design, a skill with a checklist file, testing and sharing.

Part 10 · Hands-On: Agents

Chapter 20 · Your First Subagent — two read-only reviewers, run for real and checked against the code.
Chapter 21 · Agent Teams & Parallel Work — a real parallel run, plus experimental teams (documented, not run).

Part 11 · Hands-On: Platform

Chapter 22 · Hooks Lab — six hooks tested by hand and in real sessions, including one that fails open.
Chapter 23 · MCP Lab — extend a server, test it over the protocol, register it and use it.
Chapter 24 · Plugin Packaging — one folder with skills, agents, a hook and a server, validated and proven.

Part 12 · Hands-On: Automation & Capstone

Chapter 25 · Headless Claude & CI Gates — caps, schema-checked output, load gates and a free config linter.
Chapter 26 · Capstone: Delete a Book — one real task end to end, verified independently.
Chapter 27 · Cheat Sheet & Finder — goal finder, commands, surprises and an honest ledger.

Ground Rules

How This Guide Handles Accuracy

Dated facts. The AI tooling landscape changes quickly, so every page carries an "Information current as of" date and a Sources list. Model names, prices and commands were read from vendor documentation on that date. Re-check them before relying on them.

Executed vs. illustrative. A block labelled Actual output was produced by running the command during the writing of the page, and the tool versions are recorded beside it. Anything that needs a live, authenticated Claude session and was not run is labelled Illustrative or Not executed.

No leaderboard. The guide compares tools by what they do and how they trade off. It does not publish benchmark scores or declare a winner.

LearningTree

Learning Map | Claude & AI-Assisted Development