Command-Line Interface

Most people operate a computer with a mouse and their fingers. They click on an icon, drag a file into a folder, press a button. But there's another way: you type a command as text into an empty black window and press Enter. The computer executes this command and writes the result underneath. This way of operating a computer is called a command-line interface, sometimes just called the command line or terminal. It is older than screens with icons and windows, but it is by no means extinct.

Why professionals prefer typing to clicking

A mouse-driven interface only ever shows you what someone has built in as a button. The command line shows you nothing at all, but in return allows almost anything. Anyone who knows what they want often gets there faster. A single typed command can rename a thousand files at once. With a mouse, that would be an afternoon's work.

The second big advantage is repeatability. A command is text, and text can be saved, copied, and passed on. When a developer describes how to install a program, they send three lines to copy. Instructions with mouse clicks would need twenty screenshots and would be wrong after the next update.

That is exactly why almost all servers in the world run via the command line. A server sits in a data center and often has no screen at all. You connect over the network and type. The large machines on which AI models are trained are also controlled this way.

What happens when you type a command

Between you and the operating system sits a program called a shell. The shell waits for your input and breaks it down into components. At the front usually stands the name of the desired program, followed by additional specifications. The command "ls -l pictures", for example, calls up the program ls. It lists the folder "pictures", and "-l" requests a detailed display with size and date.

Such additional specifications are called options or arguments. They are the reason the command line seems so powerful and, at the same time, so off-putting. Every program has its own options, and you have to know them or look them up. For this, there are help texts that can usually be called up with "--help".

A special trick is chaining. The output of one program can be passed directly as input to the next. To do this, you connect the commands with a vertical bar called a pipe. This turns simple building blocks into complicated processes, similar to a factory where each station passes the workpiece on.

The command line in AI tools and the news

Anyone working with AI today almost inevitably ends up in a terminal. Models are downloaded and started via typed commands. Tools like Ollama or the programs around Python work exactly this way. Codex CLI, Claude Code, and Gemini CLI are also command-line programs: you type an instruction there in plain language, and the AI then changes files in the project.

This is remarkable because it reverses an expectation. Many thought AI would make technology more colorful and clickable. In fact, it is precisely the black text window that is experiencing a comeback right now. The reason is simple: language and text go well together, and an AI can write and execute commands itself.

A common misconception, by the way, is that the command line is the same as programming. That is not true. There, you call up ready-made programs instead of writing new ones. The way back to the mouse also always remains open, since both ways of operating run on the same system and do not exclude each other.