Skip to content

Glossary

Every term defined in this syllabus, alphabetically. Each entry links back to the phase where it was introduced.


AI system with tools, decision-making ability, and multi-step execution. Unlike a chatbot, an agent can read files, run commands, and take actions. Introduced in: Phase 3 — What Are Agents

Artificial Intelligence — software that makes predictions and generates content by learning patterns from data. Not sentient. Not magic. A very sophisticated pattern matcher. Introduced in: Phase 0 — The Landscape

Application Programming Interface — how two pieces of software talk to each other. Like a restaurant menu: you place an order, get food back, no need to know how the kitchen works. Introduced in: Phase 0 — How Software Works

A specific URL where an API accepts requests (/api/users, /api/login). Introduced in: Phase 4 — Speaking Engineer

The high-level structure of how a system is organized — what pieces exist and how they connect. Introduced in: Phase 2 — Project Architecture

A record of who did what and when — critical for security and compliance. Introduced in: Phase 4 — Production Thinking

Verifying who someone is (login). Abbreviated AuthN. Introduced in: Phase 4 — Speaking Engineer

Verifying what someone is allowed to do (permissions). Abbreviated AuthZ. Introduced in: Phase 4 — Speaking Engineer

The degree to which an agent can act without human direction. Introduced in: Phase 3 — What Are Agents

What runs behind the scenes — the logic, the database, the AI calls. Introduced in: Phase 0 — How Software Works

A prioritized list of work that hasn’t been started yet. Introduced in: Phase 4 — Agentic PM

Backward Compatible {#backward-compatible}

Section titled “Backward Compatible {#backward-compatible}”

A change that works with existing code without requiring updates. Introduced in: Phase 4 — Speaking Engineer

Something preventing progress on a task. Introduced in: Phase 4 — Agentic PM

Starter code or template that gives you a foundation to build on. Introduced in: Phase 0 — Vocabulary

A parallel version of your project where you make changes safely. Like making a photocopy to draft edits on. Introduced in: Phase 2 — Git Workflow

A change that makes existing things stop working — requires users to update their code. Introduced in: Phase 4 — Speaking Engineer

Something that doesn’t work as intended. Introduced in: Phase 0 — Vocabulary

Storing frequently accessed data in fast, temporary storage. Like keeping used files on your desk instead of walking to the filing cabinet. Introduced in: Phase 4 — System Design

Content Delivery Network — servers worldwide that store copies of your content close to users. Like distribution centers in every city. Introduced in: Phase 4 — System Design

Record of what changed between versions of software. Introduced in: Phase 2 — Reading Docs

An AI interface that responds to text input but has no tools, no memory, and no ability to take actions. A conversation partner, not a worker. Compare with Agent. Introduced in: Phase 3 — What Are Agents

Continuous Integration / Continuous Deployment — automated pipeline that tests and deploys code when you push changes. Introduced in: Phase 4 — System Design

A Markdown file in your project root that Claude Code reads automatically every session. Your AI’s briefing document. Introduced in: Phase 2 — Context Engineering, Phase 3 — Harness Engineering

Command Line Interface — any tool you operate by typing text commands. See also: Terminal. Introduced in: Phase 0 — Vocabulary

Download a copy of a remote repository to your computer. Introduced in: Phase 2 — Git Workflow

Another developer examining your code before it’s merged — standard practice, not a judgment. Introduced in: Phase 4 — Speaking Engineer

A saved snapshot of your project at a specific moment. Your safety net. Introduced in: Phase 1 — Git Basics

A self-contained piece of an interface that can be reused (button, card, nav bar). Introduced in: Phase 2 — Project Architecture

Configuration — settings that control how software behaves. Introduced in: Phase 0 — Vocabulary

Where log messages and errors appear (browser console, terminal output). Introduced in: Phase 2 — Debugging

A packaged unit of software that includes everything needed to run. Like a shipping container — runs the same everywhere. Introduced in: Phase 4 — System Design

Everything the AI can “see” when answering you. Files, conversation history, instructions. Introduced in: Phase 0 — The AI Stack

Context Engineering {#context-engineering}

Section titled “Context Engineering {#context-engineering}”

Deliberately shaping what information is available to the AI so it produces better results. Introduced in: Phase 2 — Context Engineering

The size limit on how much the AI can see at once. Like a desk — bigger desk, more documents open. Introduced in: Phase 0 — The AI Stack

Percentage of code tested by automated tests. Introduced in: Phase 4 — Speaking Engineer

Where data lives permanently. Users, messages, settings. Introduced in: Phase 0 — How Software Works

Finding and fixing a bug. Introduced in: Phase 0 — Vocabulary

Assigning a task from one agent to another. Introduced in: Phase 4 — Multi-Agent Architecture

Software your project relies on that someone else built. Introduced in: Phase 0 — Vocabulary

Making your software live and available to users on the internet. Introduced in: Phase 0 — Vocabulary

Marking something as “still works but will be removed in a future version.” Introduced in: Phase 4 — Speaking Engineer

Same input always produces same output. Traditional code is deterministic. AI is not. Introduced in: Phase 3 — Vocabulary

A view showing exactly what changed between two versions. Introduced in: Phase 2 — Git Workflow

Domain Name System — translates human-readable names (google.com) to computer addresses. Like a phone book for the internet. Introduced in: Phase 1 — Deploy

The most common tool for creating and running containers. Introduced in: Phase 4 — System Design

Don’t Repeat Yourself — if you’re writing the same thing in multiple places, put it in one place and reference it. Introduced in: Phase 2 — Engineering Principles

An unusual or extreme input that exposes bugs normal inputs don’t. Introduced in: Phase 2 — Debugging

A way to represent text as numbers so a computer can measure similarity. How RAG systems find relevant documents. Introduced in: Phase 3 — Memory & RAG

The file where a program starts running. Introduced in: Phase 1 — Reading Code

Environment Variable {#environment-variable}

Section titled “Environment Variable {#environment-variable}”

A setting stored outside your code (in .env) so it can change between local and production without editing code. Introduced in: Phase 2 — Project Architecture

Text the system produces when something fails — contains diagnostic information, not punishment. Introduced in: Phase 2 — Debugging

Splitting work across multiple agents (fan-out), then combining results (fan-in). Introduced in: Phase 4 — Multi-Agent Architecture

Security system that controls what network traffic is allowed in and out. Introduced in: Phase 4 — Speaking Engineer

An option added to a command to modify behavior (e.g., --verbose). Introduced in: Phase 0 — Vocabulary

Make your own copy of someone else’s repository on GitHub. Introduced in: Phase 2 — Git Workflow

What the user sees and touches — the app, the website, the buttons. Introduced in: Phase 0 — How Software Works

AI that creates new content (text, images, code) vs. AI that classifies or sorts existing data. Introduced in: Phase 0 — The Landscape

A system that tracks every change you make to files. Like Google Docs version history for your entire project. Introduced in: Phase 0 — The Environment

A website where people store and share Git-tracked projects. Google Drive for code. Introduced in: Phase 0 — The Environment

Graceful Degradation {#graceful-degradation}

Section titled “Graceful Degradation {#graceful-degradation}”

System continues working (possibly reduced) when a component fails, rather than crashing entirely. Introduced in: Phase 4 — Multi-Agent Architecture

A constraint placed on an agent to limit its actions — a safety boundary. Introduced in: Phase 3 — Agent Patterns

When AI confidently generates something wrong — false facts, invented files, fabricated code. Introduced in: Phase 1 — First Prompt

The tool between you and the AI model. Claude Code, Cursor, ChatGPT web — these are harnesses. The harness shapes how you interact. Introduced in: Phase 0 — The AI Stack

The current position in your Git history — “you are here.” Introduced in: Phase 2 — Git Workflow

A script that runs automatically when a specific event happens. Like an alarm system. Introduced in: Phase 3 — Harness Engineering

A service that runs your project on a server so others can access it. Introduced in: Phase 1 — Deploy

Agent pauses for human approval at critical decision points. Introduced in: Phase 3 — Agent Patterns

Integrated Development Environment — a specialized text editor for code. VS Code is the most common. Like Microsoft Word, but for software. Introduced in: Phase 0 — The Environment

An operation that produces the same result whether you run it once or ten times. Safe to retry. Introduced in: Phase 4 — Multi-Agent Architecture

An unplanned event that disrupts service. Introduced in: Phase 4 — Production Thinking

When an AI model processes input and generates output. Every response = one inference call. Inference costs money. Introduced in: Phase 3 — Vocabulary

A computer’s unique address on a network. Introduced in: Phase 4 — Speaking Engineer

Refining results through multiple rounds of feedback. Rarely perfect first try — that’s normal. Introduced in: Phase 1 — First Prompt

JavaScript Object Notation — structured data format. "name": "CJ" — label left, value right. Like a fill-in-the-blank form your computer can read. Introduced in: Phase 0 — The Environment

JSON Web Token — a compact, self-contained token used for authentication. Introduced in: Phase 4 — Speaking Engineer

An organized collection of information an AI system can draw from. Introduced in: Phase 3 — Memory & RAG

Time between sending a request and getting a response — the delay. Introduced in: Phase 3 — Vocabulary

Automated checking of code style and common errors. Introduced in: Phase 4 — Speaking Engineer

Large Language Model — the type of AI behind ChatGPT, Claude, etc. Predicts the next word. Introduced in: Phase 0 — The Landscape

Distributes incoming requests across multiple servers. Like a hostess seating guests across rooms. Introduced in: Phase 4 — System Design

On your computer. Opposite of remote. Introduced in: Phase 0 — How Software Works

Your own computer acting as a server — only you can see it. Introduced in: Phase 0 — Vocabulary

Messages a program writes to record what it’s doing — your forensic trail. Introduced in: Phase 2 — Debugging

.md files — formatted text using simple codes (**bold**). Used for documentation and README files. Introduced in: Phase 0 — The Environment

Model Context Protocol — a standard for connecting AI agents to external tools and data. Like USB for AI. Introduced in: Phase 3 — MCP

A program that translates between AI and an external system, exposing tools the AI can call. Introduced in: Phase 3 — MCP

Combining changes from one branch into another. Introduced in: Phase 2 — Git Workflow

When two branches changed the same line — Git can’t decide, so you do. Introduced in: Phase 2 — Git Workflow

Architecture where the system is split into small, independent services communicating via APIs. Introduced in: Phase 2 — Project Architecture

Software between two systems that processes requests as they pass through. Like a security checkpoint. Introduced in: Phase 4 — System Design

The trained AI brain. Claude Opus, GPT-5, Gemini — different models, like different engines. Introduced in: Phase 0 — The Landscape

Continuous automated checking of system health. Introduced in: Phase 4 — Production Thinking

Architecture where everything lives in one application. Not a bad word — simpler, often the right choice. Introduced in: Phase 2 — Project Architecture

One repository containing multiple related projects. Introduced in: Phase 2 — Project Architecture

Minimum Viable Product — the smallest version that delivers value. Ship this first, improve later. Introduced in: Phase 2 — Vocabulary

Same input can produce different outputs each time. AI models are non-deterministic. Introduced in: Phase 3 — Vocabulary

Node Package Manager — installs JavaScript packages/tools from the internet. Like an app store for developer tools. Introduced in: Phase 1 — Setting Up

A standard for giving one application limited access to your account on another service, without sharing your password. Introduced in: Phase 4 — Speaking Engineer

Software whose code is public. Anyone can read, use, improve it. Introduced in: Phase 0 — How Software Works

Coordinating multiple agents or systems to accomplish a goal. Introduced in: Phase 3 — Vocabulary

A bundle of code someone else wrote that you install and use. Introduced in: Phase 0 — Vocabulary

A section within a knowledge base — like folders within a filing cabinet. Introduced in: Phase 3 — Memory & RAG

The address of a file on your computer (/Users/you/project/file.txt). Introduced in: Phase 0 — Vocabulary

The data sent with an API request. Introduced in: Phase 4 — Speaking Engineer

A sequence of processing steps where output of one becomes input of the next. Introduced in: Phase 3 — Agent Patterns

A numbered channel on your computer for network traffic. Like apartment numbers. localhost:3000 = “my computer, channel 3000.” Introduced in: Phase 1 — Vocabulary

Principle of Least Privilege {#principle-of-least-privilege}

Section titled “Principle of Least Privilege {#principle-of-least-privilege}”

Each component gets only the minimum access it needs — nothing more. Introduced in: Phase 4 — Multi-Agent Architecture

The live environment real users interact with. Abbreviated “prod.” Introduced in: Phase 4 — Speaking Engineer

The instruction you give to an AI. Introduced in: Phase 1 — First Prompt

The company that built and runs the model. Anthropic → Claude. OpenAI → GPT. Google → Gemini. Introduced in: Phase 0 — The AI Stack

Download changes from GitHub (remote) to your computer (local). Introduced in: Phase 1 — Deploy

A proposal to merge a branch, with description and review process. Abbreviated PR. Introduced in: Phase 2 — Git Workflow

Send your local Git commits up to GitHub (remote). Introduced in: Phase 1 — Deploy

A list of tasks waiting to be processed, in order. Like a line at the DMV. Introduced in: Phase 4 — System Design

When two processes try to change the same thing simultaneously, producing unpredictable results. Introduced in: Phase 4 — Multi-Agent Architecture

Retrieval-Augmented Generation — retrieving relevant info and injecting it into AI context. Like an open-book exam instead of memorizing the textbook. Introduced in: Phase 3 — Memory & RAG

Maximum number of requests allowed in a time period. Introduced in: Phase 4 — System Design

A file explaining what a project is, how to run it, and how to contribute. Introduced in: Phase 1 — Reading Code

Restructuring code without changing what it does — improving the internals. Introduced in: Phase 2 — Vocabulary

A fix that breaks something previously working. Introduced in: Phase 2 — Debugging

On someone else’s server. Opposite of local. Introduced in: Phase 0 — How Software Works

A project folder tracked by Git. Abbreviated “repo.” Introduced in: Phase 0 — The Environment

Reverting to a previous working version after a bad deploy. Introduced in: Phase 4 — Production Thinking

The top-level folder of a project. Introduced in: Phase 0 — Vocabulary

A URL path that maps to a page or action (/about, /api/users). Introduced in: Phase 2 — Project Architecture

The environment where code actually executes. Introduced in: Phase 0 — Vocabulary

An isolated environment for safe experimentation. Introduced in: Phase 4 — Speaking Engineer

System’s ability to handle growth. A restaurant that can open more tables vs. one always full. Introduced in: Phase 4 — System Design

When a project gradually grows beyond its original boundaries. Introduced in: Phase 2 — Vocabulary

A file containing commands that run automatically. Introduced in: Phase 0 — Vocabulary

Searching by meaning, not keywords. “How do I fix login?” finds “authentication errors.” Introduced in: Phase 3 — Memory & RAG

Semantic Versioning — MAJOR.MINOR.PATCH (e.g., 2.1.3). Major = breaking. Minor = features. Patch = fixes. Introduced in: Phase 4 — Speaking Engineer

A computer that’s always on, waiting to respond to requests. “The cloud” = other people’s servers. Introduced in: Phase 0 — How Software Works

Service Level Agreement — a commitment to a specific level of uptime/reliability. Introduced in: Phase 4 — Production Thinking

Specification — a document describing exactly what needs to be built and what “done” looks like. Introduced in: Phase 4 — Agentic PM

A fixed time period (usually 1-2 weeks) during which a set of tasks is completed. Introduced in: Phase 4 — Agentic PM

Secure Shell — a secure way to connect to and control a remote computer via terminal. Introduced in: Phase 4 — Speaking Engineer

Encryption that makes web traffic secure (the “S” in HTTPS). Introduced in: Phase 4 — Speaking Engineer

The combination of technologies used to build something. Introduced in: Phase 0 — Vocabulary

A list showing the chain of function calls leading to an error — read bottom to top. Introduced in: Phase 2 — Debugging

(Git) Marking which changes go into the next commit. (Infrastructure) A copy of production for testing. Introduced in: Phase 1 — Git Basics, Phase 4 — Speaking Engineer

A brief status update meeting or report. Introduced in: Phase 4 — Agentic PM

Data that changes over time and affects what the user sees (menu open? user logged in?). Introduced in: Phase 2 — Project Architecture

An agent spawned by another agent to handle a specific subtask. Introduced in: Phase 4 — Multi-Agent Architecture

Background instructions that shape how the AI behaves across an entire conversation. Introduced in: Phase 1 — First Prompt

Shortcuts taken now that create more work later. Introduced in: Phase 2 — Vocabulary

A text-based way to talk to your computer. Like texting your computer instead of tapping buttons. Introduced in: Phase 0 — The Environment

How many requests a system can handle in a given time period. Introduced in: Phase 3 — Vocabulary

The unit AI uses to measure text. Roughly 3/4 of a word. Context windows measured in tokens. Introduced in: Phase 0 — The AI Stack

The maximum tokens allocated for a task or session — controls cost. Introduced in: Phase 3 — Vocabulary

The mechanism that lets AI do things beyond generating text — reading files, running commands, searching. Also called function calling. Introduced in: Phase 3 — What Are Agents

Tests a single function or component in isolation. Introduced in: Phase 4 — Speaking Engineer

Percentage of time a system is operational. 99.9% = ~8.7 hours downtime per year. Introduced in: Phase 4 — Production Thinking

A database optimized for storing and searching embeddings. The filing system behind RAG. Introduced in: Phase 3 — Memory & RAG

Building software by describing what you want in natural language. Coined by Andrej Karpathy. Introduced in: Phase 0 — The Landscape

Virtual Private Network — encrypted tunnel between your computer and a private network. Introduced in: Phase 4 — Speaking Engineer

A URL that receives automatic notifications when something happens in another system. Like a doorbell. Introduced in: Phase 4 — System Design

The folder you’re currently in. Introduced in: Phase 1 — Git Basics

YAML Ain’t Markup Language — structured data like JSON but more readable. Used for config files. Introduced in: Phase 0 — The Environment

You Ain’t Gonna Need It — don’t build features you might need later. Build what you need now. Introduced in: Phase 2 — Engineering Principles


Total: ~150 terms. This is the shared language between vibe coders and traditional engineers. Know these, and you can hold your own in any technical conversation.