7 parts
Seven parts on Claude Code: install, the .claude directory, Skills, agents, hooks, the built-in tool inventory, and a four-phase method for trustworthy results.
Install the CLI, run a first session in a real project, and write the CLAUDE.md file Claude Code reads at the start of every turn.
Tour the .claude/ directory and the CLAUDE.md loading rules, then set repo-wide conventions for where AI-generated artifacts live.
Author your first Skills — the on-disk procedures Claude Code loads at session start — and verify both an auto-fire Skill and a slash-command Skill work.
Author your first agents — markdown files Claude Code spawns as forked sessions for focused work — and verify both fire on the right cues.
Wire hooks into the hooks stanza of .claude/settings.json so the harness fires deterministically on lifecycle events — the surface where rules-as-prose stop being advisory.
A catalogue of the tools Claude Code ships with — what each one does, when the model reaches for it, and what shows up in your session when it fires.
A four-phase method for trustworthy results — Plan, Control, Execute, Validate — and how Claude Code's native primitives compose into a reliable workflow.