Most people bounce off Claude Code at the words "open your terminal." But you're not learning software — you're learning to manage a brilliant new teammate, and you already know how to do that. This is the one-page cheat sheet for every core concept, in plain English.

Get the free Claude Code Primer cheat sheet

Claude Code isn't software you learn — it's a brilliant junior teammate you manage. Once you see it that way, every concept clicks. This free field guide (built from the video) is the whole map on a few pages, with the manager's translation beside each idea. THE WHOLE MAP, ONE LINE EACH — the agentic loop (it reads your files, makes the change, then checks its own work, and repeats); the context window (its short-term memory: your chat plus the files it opened, gone when the session ends); CLAUDE.md (a plain-English file of your rules it reads at the start of every session — the onboarding doc); memory (notes it keeps across sessions); permissions (six levels of trust); tools (read/edit files, search, run commands, browse the web, understand your code); slash commands (type /name for one exact action) vs skills (reusable playbooks it uses automatically); plan mode (it lays out the plan before touching a file, so you approve the approach instead of cleaning up a mess); subagents (focused mini-versions it spins up for big side-tasks, then reports back); connectors and MCP (one-click links to Gmail/Drive/Slack, with MCP as the advanced path); models (Haiku is fast and cheap, Sonnet is the everyday default, Opus is the deep thinker — use the nicknames, not version numbers); and undo/checkpoints (it saves a snapshot before every edit, so you can roll any change right back). THE ONE TABLE TO MEMORIZE — the 6 permission modes and what each does without asking: Default (asks before anything), Accept edits (edits files on its own), Plan (looks but never touches), Auto (most things with a background safety check), Don't ask (only your pre-approved actions, great for scripts), and Bypass (skips every check — throwaway test machines only, never your real computer). Tap Shift+Tab to dial the trust up or down anytime. PLUS the 3 beginner mistakes to skip (a vague brief, no CLAUDE.md, and not checking the work) and a ready-to-paste starter CLAUDE.md you can drop into your own project today. The one-line takeaway: you don't need to be technical — you just need to be a good boss.

Subscribe to Hyperautomation AI ReportGet the PDF freeKeyword: PRIMER

Free. No spam. Unsubscribe anytime.