📖 Jargon Library

Plain-English definitions for every technical term in this course. You can also hover over any dotted word in any module page for an instant definition.

🔤 API

Application Programming Interface — a way for different software programs to talk to each other. Think of it as a waiter: you (the program) tell the waiter (API) what you want, and it brings it from the kitchen (the other program). The Claude API lets your scripts call Claude directly.

🤖 Agent / Agentic AI

An AI that works autonomously across multiple steps to complete a goal, rather than just answering one question. It can use tools, run code, read files, make decisions, and fix its own errors. Claude Code operates agentically when given complex tasks.

🔑 API key

A long secret code (like a password) that lets software connect to Anthropic's servers. It looks like "sk-ant-api03-...". Keep it secret — anyone with your API key can use your Claude account at your expense.

💻 bash

A specific type of shell (command interpreter) used on Mac and Linux. When you open the Terminal on a Mac, you're usually using bash (or zsh, which is very similar). It's the language you type commands in.

🌿 branch

In Git, a parallel version of your project where you can make experimental changes without affecting the main version. Like working on a copy of a document while keeping the original safe.

📋 CLAUDE.md

A special instruction file you create in your project folder. Claude Code reads it automatically at the start of every session, so it already knows your project, conventions, and preferences. Like a permanent briefing document for your AI assistant.

💻 CLI (Command Line Interface)

A text-based way to give instructions to your computer by typing commands, rather than clicking on buttons. The Terminal is a CLI. Claude Code is a CLI tool — you interact with it by typing.

📸 commit

In Git, a saved snapshot of all your changes at a specific moment. Like pressing "Save" but smarter — each commit has a message explaining what changed and why. You can always go back to any previous commit.

🧩 context window

The maximum amount of text Claude can "see" at one time — including your conversation, any files it's read, and instructions. Once the context window fills up, older content is forgotten. Think of it as Claude's working memory. Use /compact to clear space.

💾 dependency

Software that your project needs in order to work. Like how a Stata do-file might require the "outreg2" package to be installed first. If a dependency is missing, the code won't run.

📝 diff

Short for "difference" — a visual display showing exactly what changed between two versions of a file. Added lines appear in green (+), removed lines in red (-). When Claude proposes a file edit, it shows you a diff so you can review before approving.

📁 directory

Another word for "folder". A directory contains files and can contain other directories (sub-folders). When people say "navigate to a directory", they mean "go to that folder in your terminal".

📄 do file

In Stata, a .do file is a script containing Stata commands — like a recipe of analysis steps that can be run automatically.

🐳 Docker

A tool for packaging software with all its dependencies into a "container", so it runs the same way on any computer. Useful for reproducible research environments.

🌍 environment variable

A named value stored in your computer's system settings (not in your code). Used to store secret keys, settings, and configuration. For example: ANTHROPIC_API_KEY stores your Claude API key safely without putting it in your code files.

🌐 Git

A version control system that tracks every change to your files over time, so you can see what changed, who changed it, and undo any mistake. Essential for research reproducibility and for working safely with Claude Code.

🐙 GitHub

A website that hosts Git repositories online. Free for public projects. Used to share code, collaborate with co-authors, and back up your work. GitHub Pages hosts static websites (like this course!) for free.

🔍 glob pattern

A shorthand way to describe multiple files using wildcards. * means "any text". Examples: *.do means all Stata do-files; data/*.csv means all CSV files in the data folder; **/*.R means all R files anywhere in the project.

🪝 hook

An automated action that runs automatically when something specific happens in Claude Code. For example: "every time Claude saves a file, run a linter to check for errors". You configure hooks once and they run silently in the background.

💻 IDE

Integrated Development Environment — software designed for writing code. Has helpful features like autocomplete, error highlighting, and file browsing. Examples: VS Code (free), RStudio (for R), Stata's do-file editor.

📊 JSON

JavaScript Object Notation — a simple text format for storing structured data. Looks like a dictionary: {"name": "value", "number": 42}. Claude Code's settings files are written in JSON format.

📓 Jupyter notebook

An interactive document (file extension .ipynb) that mixes explanatory text, code cells you can run, and their outputs. Very popular in Python for data science. Can be run locally or via Google Colab in a browser.

📝 LaTeX

A document preparation system (pronounced "lah-tech") used in academia, especially economics and mathematics. You write text mixed with formatting commands (like \textbf{bold}), and it produces a beautifully typeset PDF. Standard for journal submissions in many fields.

🤖 LLM (Large Language Model)

The type of AI that powers Claude. It's a neural network trained on enormous amounts of text to understand and generate human language. "Large" refers to the number of parameters (billions) that the model learns during training.

🔌 MCP (Model Context Protocol)

A standard way to connect Claude Code to external tools and data sources — databases, websites, GitHub, Zotero, etc. An MCP server is a small program that gives Claude new capabilities, like a plugin. Configured in settings.json.

📦 Markdown

A simple text formatting system where you use symbols to indicate formatting: **bold**, *italic*, # Heading 1, ## Heading 2, - bullet point. CLAUDE.md and README files are written in Markdown. GitHub displays it beautifully.

⚠️ merge conflict

In Git, when two people edited the same lines of a file differently and Git can't automatically combine them. You have to manually decide which version to keep. Claude Code can help you resolve merge conflicts.

📦 npm

Node Package Manager — a tool for downloading and installing JavaScript software. It's like Stata's "ssc install" or R's "install.packages()". You use it to install Claude Code with: npm install -g @anthropic-ai/claude-code

🟢 Node.js

A JavaScript runtime environment — software that lets JavaScript programs run on your computer (outside a browser). Claude Code is built with Node.js, so you need to install it before Claude Code.

🔓 open source

Software whose source code is publicly available for anyone to read, use, modify, and share. Free to use. This course is open source — you can see all the code on GitHub and contribute improvements.

🔐 permissions

Settings that control what Claude Code can do automatically vs. what requires your explicit approval. You can set some actions to "always allow" (e.g., editing .do files) and others to "always ask" (e.g., deleting files).

💬 prompt

The message or instruction you type to Claude. The quality of your prompt directly affects Claude's output. A good prompt includes context (what the project is), the task (what you want), constraints (rules to follow), and desired output format.

💻 PowerShell

A command-line shell for Windows — a more powerful version of the Command Prompt. You can use it to run Claude Code on Windows, though WSL (Linux environment) is often recommended instead.

📁 README

A file (usually README.md) at the top of a project folder that explains what the project is, how to install it, and how to use it. Claude Code can read your README as part of understanding your project, and can write one for you.

📊 R script

A .R file containing R commands for data analysis, cleaning, or visualization. The R equivalent of a Stata .do file.

📊 regression

In statistics, a method to estimate the relationship between variables — one of the most common tools in applied economics and social science research.

📁 repository (repo)

A project folder tracked by Git, usually also hosted on GitHub. Contains all your project files plus the full history of every change ever made. When someone says "the repo", they mean the Git-tracked project folder.

🔍 regex (regular expression)

A pattern language for searching and matching text. Powerful but complex. Example: \d{4} matches any 4 consecutive digits. You can ask Claude Code to write regex patterns for you — just describe what you want to match in plain English.

🔒 SSH

Secure Shell — a way to securely connect to and control another computer (like a university server) over the internet, using the terminal. Often used when doing computation on a university high-performance computing cluster.

🖥️ shell

The program that interprets your typed commands in the terminal. Common shells: bash and zsh (Mac/Linux), PowerShell (Windows). When you type "ls" or "cd", the shell interprets and executes that command.

⚡ slash command

A shortcut command starting with / that you type in Claude Code. Built-in examples: /help, /clear, /exit. You can also create custom slash commands (called "skills") for your own repeated tasks.

📊 Stata

A commercial statistical software package widely used in economics, public health, and social science. Known for its strong panel data and survey data tools. Uses .do files (scripts) and .dta files (data).

⚡ skill

A custom slash command you create for Claude Code. It's a Markdown file with instructions stored in .claude/commands/. When you type /skill-name, Claude reads those instructions and follows them. Like a macro or template for AI tasks.

📤 stdout

Standard output — the default place a program sends its text output (usually your terminal screen).

⚠️ stderr

Standard error — where programs send error messages (also usually your terminal, but separate from stdout so errors can be handled differently).

🖥️ terminal

A program that gives you a text-based window to type commands to your computer. On Mac: called "Terminal" (in Applications → Utilities). On Windows: "Command Prompt", "PowerShell", or "Ubuntu" (WSL). On Linux: "Terminal" or "Konsole".

🪙 token

The basic unit Claude uses to process text — roughly ¾ of a word (about 4 characters). API pricing is measured per token. "1,000 tokens" ≈ 750 words ≈ 1.5 pages of text. Long conversations use more tokens and cost more.

💻 VS Code

Visual Studio Code — a free, popular code editor made by Microsoft. Works with almost every programming language. Has many extensions. Claude Code integrates with VS Code so you can see file changes directly in your editor.

🐧 WSL (Windows Subsystem for Linux)

A built-in Windows feature that lets you run a Linux environment inside Windows. Type "wsl --install" in PowerShell to set it up. Recommended for running Claude Code on Windows, as many developer tools work better in Linux.

📄 YAML

Yet Another Markup Language — a text format for configuration files, simpler than JSON. Uses indentation to show structure. The _config.yml file for GitHub Pages uses YAML. Claude Code can read and write YAML files.