Claude Code in Practice: CLAUDE.md, permissions, subagents, skills, hooks and MCP, with templates

Installed Claude Code and using it as a chat that edits files? This guide shows you the rest, in direct, practical English.

10 hands-on chapters, in the order worth setting things up:

- a short CLAUDE.md that Claude actually uses;

- permissions and security: what runs without asking, what never runs and where your secrets live;

- subagents with a defined role and limited tools, such as a reviewer without the editing tools;

- skills and slash commands for the procedures you repeat;

- hooks that format, run the tests and block dangerous commands without relying on the model's memory;

- MCP servers for GitHub, the browser and databases, with environment variables instead of tokens;

- automation with claude -p in scripts, in CI and on a schedule;

- the routine of a small agency serving several clients;

- troubleshooting, with a table of symptoms and causes.

It comes with a template pack with English comments (a few file names keep their original Portuguese names): CLAUDE.md, settings.json with permissions and three Node hooks (written to run the same on Windows, macOS and Linux), three subagents, two skills, .mcp.json and a piece-by-piece installation guide.

Every feature, command and setting was checked against the Claude Code documentation (code.claude.com) on October 27, 2026, and the book lists the pages consulted. Claude Code changes often, so check the version you use.

For developers, freelancers and small teams who already use Claude Code and want to get more out of it, safely. The examples use Node/TypeScript projects, Python scripts and a small agency's day-to-day work.

Independent product, written by someone who uses Claude Code every day. Not affiliated with Anthropic. Claude and Claude Code are trademarks of Anthropic.

Format: 86-page PDF (A5) + template pack (a PDF with every file and the .zip attached).