Terminal-based

Most programs on a computer are operated with the mouse. You click on icons, open menus, and drag windows across the screen. But there is a second, much older way of operating a computer: you type a command into a plain text window and press Enter. The computer responds likewise with text. Such programs are called terminal-based, because this text window is called a terminal. The name comes from a time when users sat at simple screen devices — precisely terminals — that were connected to a large computer in the next room.

Why professionals stuck with the text window

Typed commands seem old-fashioned, but they are often faster than clicking. A single command can rename a thousand files at once. With the mouse, that would mean a thousand individual clicks. Anyone who knows what they want gets much further with text.

The more important reason is repeatability. A command is text, and text can be saved, passed on, and made to run automatically. This is how scripts arise, that is, lists of commands that a computer works through without a human. Mouse clicks cannot be written down in this way. That is why practically all automation in data centers runs via commands rather than windows.

On top of that, there is a very practical point: servers usually have no screen at all. You connect to them over the network and get only a text window. Anyone running software on remote machines therefore often has no choice. A text window also requires hardly any data transfer and works even on a slow connection.

Command, arguments, and the text streams behind them

A command consists of a name and additions. The name says which program should start. The additions say what it should refer to and how it should work. If, for example, you type the name of a copy program followed by two file names, the program knows what to copy where. The program that reads and executes this line is called a shell.

Every such program has three text channels: one for input, one for results, and one for error messages. That sounds technical, but it has a nice consequence. You can pipe the result of one program directly into the input of the next. This chaining is called a pipe. This way, you build large workflows out of small tools without writing a single line of your own software.

An important distinction: terminal-based does not necessarily mean bare command line. There are also programs that draw entire interfaces with frames, lists, and colors in the text window. These are called text interfaces, or TUI for short. They can be operated like a windowed program, but still consist only of characters.

From coding assistant to error message in the studio

In the recent AI news, the term mainly comes up in connection with programming assistants. Tools such as Claude Code, Codex CLI, or Gemini CLI deliberately run in the terminal rather than as their own window. You describe in plain language what should be changed, and the assistant edits the files directly in the project folder. The advantage: it sits exactly where the developers already work and can run the same commands they do.

You also encounter this when trying out AI yourself. Anyone who wants to run a language model on their own machine usually types a command to do so. Guides online therefore often show a line that you are supposed to copy and paste. That is exactly what is meant when it says somewhere that a tool is only available terminal-based.

A common misconception is that the terminal is a sign of backwardness. In fact, it is a deliberate choice. Text commands are stable, well documentable, and can be remotely controlled by other programs. Graphical interfaces often emerge only later, as a convenient shell wrapped around an already existing terminal-based tool.