Learn / Glossary
Claude Code & Codex glossary
If you are learning to build with an AI coding agent and keep hitting words you do not know, start here. These are the terms that come up most when non-technical people use Claude Code or Codex, explained in plain English. No jargon, no assumed knowledge.
The basics
- AI agent
- Software that can take actions for you, not just chat. An agent can read your files, run commands, and make changes. Claude Code and Codex are agents.
- Large language model (LLM)
- The AI that reads and writes text. Claude and GPT are LLMs. It is the engine doing the actual thinking.
- Model
- A specific version of the AI, like Claude Opus or GPT-5. Newer or larger models are usually more capable, sometimes slower or more expensive.
- Token
- The unit the AI reads and writes in, roughly three quarters of a word. Usage is billed by tokens, and context limits are measured in them.
- Context window
- How much text the AI can hold in mind at once: your files, the conversation, and your instructions. When it fills up, older detail gets summarised or dropped.
- Prompt
- What you type to the AI. The clearer your instruction, the better the result.
The tools
- Claude Code
- Anthropic's coding agent. It runs in your terminal, desktop app, or code editor, and can read, write, and run your project.
- Codex
- OpenAI's coding agent. Same idea as Claude Code, built by a different company on a different model.
- CLI (command-line interface)
- A way to control your computer by typing commands instead of clicking. This is where Claude Code lives.
- Terminal
- The app on your computer where you type commands. On a Mac it is called Terminal, on Windows it is PowerShell or Command Prompt.
- IDE
- A code editor such as VS Code or Cursor. Claude Code offers extensions that run it directly inside these editors.
Working with the agent
- Slash command
- A shortcut you type starting with a forward slash, like /review, to trigger a saved action without re-explaining it.
- Skill
- A reusable package of instructions that teaches the agent to do one thing well. Install it once and reuse it anytime. See the essential skills list below.
- Subagent
- A helper the main agent spawns to handle a focused task, such as research, so the main conversation stays clean and on track.
- MCP (Model Context Protocol)
- An open standard that lets the AI connect to outside tools and data, like your calendar, a database, or LinkedIn. Think of it as a universal adapter.
- Hook
- An automated action that runs in response to an event, such as running tests after every edit. You set it once and it runs itself.
- CLAUDE.md
- A plain-text file in your project where you write standing instructions for Claude Code. It reads this file at the start of every session.
- Memory
- Notes the agent keeps between sessions so it remembers your preferences and context instead of starting fresh each time.
- Plan mode
- A mode where the agent proposes a plan and waits for your approval before changing anything. Useful for bigger or riskier tasks.
Code and shipping
- Repository (repo)
- A folder that holds your project's files along with the full history of every change.
- Git
- The system that tracks every change to your files, so you can undo, branch, and work with others without losing anything.
- Commit
- A saved snapshot of your changes with a short note describing them. Your project history is a series of commits.
- Push
- Sending your committed changes up to a shared server such as GitHub, so they live somewhere other than your laptop.
- GitHub
- A website that stores repos online, lets people collaborate, and can run automation when you push changes.
- Branch
- A parallel copy of your project where you can make changes safely before combining them back into the main version.
- Deploy
- Publishing your site or app so other people can use it live on the internet.
- Build
- The step that turns your source files into the final files a browser or server actually runs.
- CI/CD
- Automation that builds, tests, and deploys your project every time you push changes. Hands-off shipping once it is set up.
- Dependency
- Pre-written code your project relies on, installed rather than written from scratch. Also called a package or library.
- API
- A way for two pieces of software to talk to each other through a defined set of requests.
- Environment variable
- A setting stored outside your code, often a secret like a password or key, so it stays private and out of your files.
- Frontend and backend
- Frontend is what users see and click, the interface. Backend is the behind-the-scenes logic, data, and servers.
- Localhost
- Your own computer acting as a private server, so you can preview a site before it goes live for everyone.
Know the words, now build the thing. I run hands-on workshops where non-technical people build and deploy a real website with Claude Code in one day.