Terminal or app: teachers compare ways of working with a coding agent

A conversation among users of Claude and other coding agents set two ways of working against each other. Those who prefer the terminal describe it as direct communication with the AI, with no buttons or icons in between, and use it even to organise personal documents and maintain a memory system; they keep the web interface for quick queries, mostly from their phone.

Those who prefer the desktop app argue that it is the same thing with a graphical interface, that it starts in a second, allows several chats at once within a single project, organises shortcuts and makes it easier to connect external tools. They add that, when working by projects and chats, everything depends on the service's own memory. Someone pointed out that working with information in the cloud raises the question of privacy, and another person replied that their method keeps the documents on their own computer with encrypted backups.

There was agreement on the essentials: web, app and terminal are three routes to a similar result, and knowing how to choose the tool and the interface for each case is part of learning, something that some members of the group linked to AI literacy. It is an exchange of preferences and experiences, with no verifiable data.