Stage 2 of 10 · 6 lessons · ~1 week

Setup and project context

Install your tool in a real repository, write a short instructions file it loads every session, add scoped rules, keep secrets out of its reach and learn to point it at exactly the right files.

Start: Installing and starting Claude Code

Lessons· one lesson for each topic, in this order

Do this

Install your tool and start it from the root of a real project.
Draft the instructions file with the built-in command, then cut it to the lines that change behaviour.
Add one scoped rule for a folder or file type and check it loads.
Block .env and key files from the tool, and confirm .env is in .gitignore.
Do one small task using precise file references instead of a description.
Commit the instructions and rules so your future self and teammates get them.

Start with

CourseFree
Claude Code 101
Anthropic · Beginner
The installation, first prompt and CLAUDE.md lessons match this stage

Keep handy

Claude Code Documentation
The memory, settings and permissions pages cover every file named in this stage
AGENTS.md
The shared format your tool can read, with examples of what to put in it

Also worth your time

How Claude remembers your project (opens in a new tab)Read
CLAUDE.md locations, imports, .claude/rules and how AGENTS.md is read
Effective Context Engineering for AI AgentsGo deeper
Why context is the scarce resource for agents, and how to curate it

Before moving on: Claude Code running from your project root, a committed CLAUDE.md under 200 lines that you have tested, one rule in .claude/rules with a paths pattern, deny rules for .env and keys in .claude/settings.json, and a first task done with @ file references.

Tauseef's advice

Treat the instructions file like code: short, accurate and reviewed. Every time the tool makes the same mistake twice, add one line; every time a line stops mattering, delete it. And set up secret handling before the first session, not after the first leak.