CLI

Most people operate a computer through images and clicks. You see windows, folder icons and buttons, and at most type a file name. But there’s another way: you open a plain text window and type a command into it. The computer executes it and responds with text as well. This way of operating a computer is called CLI, spelled out as Command Line Interface, or in German 'Kommandozeile'. It is not an outdated stopgap, but for many tasks still the fastest method to this day.

Why professionals prefer typing over clicking

The most important advantage is repeatability. A mouse click disappears the moment you make it. A typed command, on the other hand, is text and can be copied, saved and passed on. Anyone who has to perform a task a hundred times writes the commands into a file and lets them run automatically. This is exactly what a large part of modern software development is based on.

Then there’s precision. A guide with screenshots becomes outdated as soon as someone redesigns the interface. A command remains the same command. That’s why installation instructions for developer tools almost always consist of lines to copy. Error messages are also easier to share this way, because you can simply paste them as text into a forum.

A third reason is distance. Servers in data centers have no screen and no mouse. You connect to them over the network and control them exclusively with text commands. Anyone who operates cloud services today works almost exclusively this way.

The structure of a command

A command usually consists of three parts. First comes the name of the program, for example git, a tool for version control. This is often followed by a subcommand that says what the program should do. At the end come additional details, such as a file name or an option like --help. Options are recognized by the dashes placed in front of them.

The program that receives the input is called a shell. It reads the line, looks for the matching program and starts it. It then displays the output and waits for the next input. The small piece of text at the start of the line, the so-called prompt, usually reveals which folder you are currently in.

This becomes especially powerful through chaining commands together. The output of one program can become the input of the next directly. This allows small tools to be combined into larger workflows without anyone having to write a new program. The price for this is the learning curve: you have to know the names of the commands, because none of it is visibly displayed on the screen.

Command lines in AI tools

Almost every developer tool today comes with its own command line. It’s used to launch programming environments, install additional packages, and upload code to shared project repositories. The major cloud providers such as Amazon, Microsoft and Google also supply their own CLIs, which are used to create and shut down servers.

In the AI industry, the term has recently appeared frequently in the news. Providers such as OpenAI, Anthropic and Google have released coding assistants that run directly in the terminal. There, you describe in plain language what should be changed, and the model edits the files itself. The reason for this design is practical: the command line already has access to all the tools of a project.

CLI should not be confused with API. A CLI is meant for humans who type commands. An API is an interface through which programs communicate with one another. Often, an API is exactly what lies behind a CLI, just with a keyboard-input interface placed in front of it.