Plain-English terms you might bump into. Scan when a word throws you.
/clear
Start a fresh conversation — the AI forgets the current chat. Your
saved progress in progress.md is safe, and the old
conversation stays available under /resume. See also
Compacting.
/compact
Squash a long conversation into a short summary so the AI can keep going without running out of memory. You keep the thread of what you were doing, but trade the fine detail for the summary. Handy when a session is getting long but you're mid-task. See Compacting below.
/context
Show how full the conversation's memory (the context window) is, and what's taking up the room. Useful when things feel slow and you want to see whether it's time to clear or compact.
/exit
Leave the tutor cleanly. You can also just close the terminal window; your progress is saved as you go, so nothing is lost.
/help
Show the list of all available slash commands the tutor knows.
"undo that"
To undo the last AI change to your files, just ask the tutor in chat: "undo that" or "take us back to the last save-point." The tutor handles the rollback. There's no special command to memorise.
Agent
An AI that doesn't just answer questions but can take actions for you — reading and changing files, running commands, fetching things from the internet — to carry a task through from start to finish. Your tutor is an example: Claude working as an agent. (A plain chatbot just talks; an agent does.)
AI
Short for "artificial intelligence." Day to day, it means a computer system that can understand and produce language, answer questions, and help with tasks — like Claude. The thinking part inside it is the AI model (below).
AI model
The "brain" of the AI: the part that does the actual thinking. Different companies have their own families: Claude (Anthropic, the one your tutor uses), GPT (OpenAI, the one inside ChatGPT), and Gemini (Google). New versions come out every few months; bigger version numbers are newer.
Alias (launcher)
A short word you teach your terminal that runs a longer command
for you. tutor is one: it opens Claude in your learning
folder. You can have one per project. Type alias on its
own in the terminal to list the ones you have.
Bookend (step-2 / step-10a bookend)
The tiny thing you make at step 2 doesn't get retired — you keep improving it as you learn. At step 10a the tutor digs the very first version out of your save-point history and puts it next to what it's become, with your day-one prompts read back alongside. The distance between the two is how far you've come.
Cage
The sealed-off room your tutor runs in: a small Linux environment
living inside your computer. From inside it, the tutor can only see
the FLT folder — your photos, emails, passwords and
everything else are behind the wall, not by rule but by how the room
is built. You can see when you're inside it: the prompt shows an
[FLT cage] tag. Installed from the
Getting started page; driven day-to-day by
the flt command.
Cageless setup
The tutor installed directly on your computer, with no cage around it. It teaches the same curriculum; the difference is that the AI's boundary is a rule it follows rather than a wall around it. The standard way in on Linux, and the fallback for Windows and Mac computers that can't host the cage. If you think you need it, message Becky.
cd
"Change directory": moves you into a different folder in the
terminal. For example, cd ~/Documents/my-game takes you
into the my-game folder. cd .. takes you back up one
level.
Claude
Anthropic's AI model family, the one your tutor uses. Comes in tiers: Opus (the most capable, slower, pricier), Sonnet (middle ground), and Haiku (smallest and fastest). You'll often see version numbers attached (Claude Sonnet 4.6, Opus 4.8); higher numbers are newer.
Claude Code
The Anthropic tool that runs the tutor in your terminal. "Code" in the name nods at the kind of work it's best at; "CLI" (the thing it is) stands for "command-line interface": you talk to it by typing.
CLAUDE.md
A text file that tells Claude how to behave. Your tutor's lives
inside your project folder at
~/Documents/my-claude-project/CLAUDE.md. Claude reads
it every time you launch the tutor. Don't edit it:
if you change it, the tutor breaks. In future projects you can add
your own CLAUDE.md to each project folder to customise
Claude's behaviour there; the tutor's closing lesson walks you
through how to ask for one.
Command
Text you type to make something happen. Some commands go in the
terminal (like claude), some go in the tutor and start
with / (like /help).
Command line developer tools
A free toolkit from Apple that a Mac needs before git
works. You don't go looking for it: the first time you type
git --version into Terminal, your Mac offers to install
it. Windows and Linux don't have this; there you install Git
directly.
Compacting
When a long conversation is squashed into a short summary so the AI
can keep going (you can trigger it with /compact). You keep
the gist of what you were doing; you lose the fine detail. The
alternative is to start fresh, which is sharper but the AI remembers
nothing. The Hints and Tips page
explains when to use which.
Context window
The AI's memory limit for a single conversation — how much it can hold in mind at once. Everything you type and everything it replies takes up room (measured in tokens). When it fills up, the conversation slows down and the AI starts forgetting the beginning. The Hints and Tips page explains how to manage it.
Continue learning
What you type to the tutor to pick up where you left off, after you've reopened it for a new session. The tutor reads your progress and carries on.
Ctrl+C
The terminal's "cancel" key. At the terminal prompt it interrupts whatever's running. Inside your tutor, one press clears what you've typed; two presses in a row quit. It does not copy text; see the install page for the copy shortcut on your system.
curl
A little tool, built into your terminal, that fetches things from the internet. The setup command in step 5 of the install guide uses it to download your tutor.
End lesson
What you type to the tutor when you want to stop for the day. The tutor saves your progress, gives you your pick-up-next-time reminder, and tells you it's safe to close the window. A full stop or other punctuation on the end is fine: "End lesson." works just the same.
Esc
The "stop" key inside your tutor. One press tells the AI to stop what it's doing in the middle of a response. Press it twice quickly to wipe what you've typed and start over.
flt
The one command that drives the cage from outside it. On its own,
flt opens the cage; flt tutor opens it and
starts the tutor in one go (the everyday tutor command is
a shortcut for exactly this); flt restart restarts
it, flt reset wipes and rebuilds it, flt update
--tutor refreshes the tutor content, and flt --help
explains everything. Only exists on cage installs — the cageless setup
doesn't have or need it.
Folder
A container for files on your computer — the same folders you already see in File Explorer (Windows), Finder (Mac) or Files (Linux). Sometimes called a directory.
Gemini
Google's AI model family. Different from the one your tutor uses (which is Claude). Mentioned here because you might hear about it.
Git
A tool that quietly tracks every change to your files so you can go back to any earlier version. The tutor handles the git work for you.
GitHub
A website that stores your project folders online. When you "push", a copy of your folder (with its whole save-point history) goes up to the site; the copy on your computer stays put. Only you can see your work there, unless you invite someone or choose to make it public. Think Google Drive or iCloud backup, but for project folders.
GPT
OpenAI's AI model family, the one inside ChatGPT. Different versions like GPT-4 and GPT-5; the "mini" and "nano" variants are smaller and faster but less capable.
Hook
A small automation that runs by itself whenever a particular thing happens — for example, playing a sound every time Claude finishes its turn. You ask Claude to set one up once, and it works every time after. A "set it and forget it" helper.
Install
Put a piece of software onto your computer so you can use it.
ls
"List": shows everything in your current folder. Handy when you can't remember what a folder or repo is called and want to see your options.
LTS
"Long-Term Support": the stable version of software, recommended for most people. You'll see it on the Node.js download page.
Node.js
The engine Claude Code runs on. Has to be installed before Claude Code itself.
npm
Node.js's installer for adding extra tools. The tutor's setup uses it once, to install Claude Code.
Permission mode
The setting that controls when Claude asks you before doing things. Yours is set to "auto": no confirmations pop up while you work, but every action is checked in the background against what you actually asked for, and anything that doesn't fit gets blocked. The tutor also pauses in chat before any meaningful change so you can review it there. Later, you can change this if you want more — or less — checking. See the Going further page for the options.
Plan file (plan.md)
A file where you and the AI write down what you're building before building it: what the project is, what files there'll be, what order to do things in. It matters because the AI forgets everything when you clear the conversation — a plan it can re-read is how a fresh AI knows what it's meant to be building. (This is step 8 of the curriculum; see also starting your own projects.)
progress.md (the tutor's memory)
A file the tutor keeps in your project folder, at
~/Documents/my-claude-project/progress.md. It remembers
where you got to. You can open it in Notepad (Windows), TextEdit
(Mac) or Text Editor (Linux); no magic, just a text file.
Project folder
The single folder where your learning lives, at
~/Documents/my-claude-project. The tutor creates it on
day one. For now, everything goes here.
Prompt
What you type to the AI to tell it what you want.
Push
Sending a copy of your save-points up to GitHub. A save-point (commit) lives only on your computer until you push it. Your local copy stays put; pushing just puts a copy online.
Repo (repository)
A folder that git is tracking changes for. Your project folder is a repo. ("Repo" is just short for "repository"; same thing.)
Resume kit
The phrases for getting back into the tutor after a break (open
your terminal, type flt, then tutor, then
type Continue learning). The Tutor Notes page has the
full text.
Save-point
A snapshot of your work you can return to if something goes wrong. Git makes these for you (the technical name is a "commit"). The tutor takes care of making save-points as you work.
Session log
The running list of past sessions inside progress.md:
what was covered, how it went, anything to remember next time. The
tutor writes a fresh entry at the end of each session.
Slash command
A command you type to the tutor that starts with
/: like /help or /clear. The
tutor recognises these and acts on them instead of treating them as
a normal message. You hardly ever need them; the tutor handles
things in chat.
Status line
A strip of information along the bottom of the Claude window. You can have it show things like how full your context window is, as a percentage, so you can keep an eye on it at a glance. Claude can set one up for you.
Tab
The autocomplete key in the terminal. Start typing a folder or filename, press Tab, and the terminal finishes the name for you. Saves typos.
Tangent tolerance
Whether you'd rather the tutor stick strictly to the path it's teaching, or feel free to mention interesting side-paths when they come up. You answer this in the session 1 diagnostic; the tutor respects your choice.
Terminal
A text window where you type commands instead of clicking. On Windows we always use the built-in PowerShell (press the Windows key, type powershell, Enter). Windows does have other terminal programs — you may hear of Command Prompt or Windows Terminal — but every instruction on this site says PowerShell, so that everyone's screen matches the instructions and the help. On a Mac it's the built-in app actually called Terminal (find it with ⌘ Cmd+Space); on Linux it's usually also called Terminal (look in your applications menu, or try Ctrl+Alt+T).
Thinking mode (effort)
Some AI models can be asked to "think harder" before answering; they take longer but reason more carefully. You'll hear it called different things: extended thinking (Claude), reasoning (GPT), deep think (Gemini). The trade-off is always the same: slower in exchange for more thorough. Your tutor uses Claude's normal mode for now; that's plenty for learning.
Token
The unit an AI measures text in — roughly a word, or a bit of a word. Both what you type and what the AI replies are counted in tokens, and they're what fill up the context window. You never have to count them yourself; just know that "more tokens" means "more memory used."
Version (of the tutor)
Which edition of the tutor's teaching files you have on your computer. The tutor keeps improving after you install it, so yours can fall behind. Ask the tutor "What version are you?" to check, and see updating your tutor to get the latest; updating never touches your work or progress.
WSL2
"Windows Subsystem for Linux": a feature built into Windows, made
by Microsoft, that runs a real Linux environment inside your PC. It's
what the Windows version of the cage is built on.
You never work with WSL2 directly — the flt command
handles it for you.