Why the terminal takes on longer jobs than apps

Following a single task to the end (comparing quotes from three suppliers and drafting a reply), the write-up explains why this kind of work does not exist as a button in an email inbox or a spreadsheet program. The diagnosis is that the work people want to do is longer than the functional unit of any one app, which is why a command-line environment is used, where installed programs and connected services can be combined as needed. On top of this, it defines the harness as the surrounding setup that attaches tools, an execution environment, permission checks and work logs to the model, and places MCP and skills along the same line. Especially practical are the division of labor in which the executed program handles calculation while the model only judges which values to compare and by what criteria, and the principle of putting an approval step before actions that are hard to undo. However, there are no screens of it running and no measurements, so you cannot get as far as how to set it up.

This is a note: my own write-up after reading several sources. It has no source link; events checked against a source are kept as records.