Anthropic Releases a Tool That Writes and Runs Code From the Command Line

Claude Code lets a programmer describe a change in plain English; the program edits the files, runs the tests, fixes what breaks, and reports back. It's aimed at software developers and marks a shift from AI that suggests code to AI that does the work.

Anthropic released a tool today called Claude Code. It runs in the terminal — the plain text window programmers use to control a computer directly — and it takes instructions in ordinary English. Tell it to add a feature, fix a bug, or reorganize a project, and it reads the relevant files, makes the changes across as many of them as needed, runs the tests to check its work, fixes what it broke, and tells you what it did. You review the result the way you'd review a colleague's.

This is a step beyond the coding assistants that came before. GitHub Copilot, released in 2021, suggested the next line as you typed. Claude Code is given a goal and left to accomplish it, touching many files and running commands on its own. It is the computer-using idea from last October, aimed squarely at the job its own makers know best.

It is for programmers, not the public, and it costs money by the amount of work it does rather than a flat fee. Early users describe it as genuinely capable on real projects and also as something that needs watching — it can go down a wrong path confidently, and it runs up a bill while it works.

The reason it matters beyond software is what it signals. Programming is the task these companies understand most deeply and can measure most precisely, so it is where the shift from advice to action shows up first. A tool that can be handed a real engineering task and complete it is the clearest sign yet of where the rest is going.