The IDE is built around managing code and making it easier to write. File trees, autocomplete, refactoring shortcuts. A whole workspace organized around the person typing the next line.
When AI handles that work, I think it is time to let the IDE go.
You could keep it "just in case." It does not hurt to have it around, right?
On October 29, 2026, Cursor disclosed that its marketplace had distributed malicious copies of the Claude Code extension, versions 2.1.312 and 2.1.313. They attempted to install a persistent SSH backdoor. Microsoft's analysis confirmed the malicious packages were not distributed through Microsoft's marketplace or the official npm package. Successful remote access depended on the machine's permissions and configuration; the attempt did not succeed on every machine.
Hours after those malicious extensions started circulating, Binance suffered the largest crypto heist recorded to date, with roughly $6.3 billion stolen. Its initial investigation pointed to a possibly compromised DevOps engineer's laptop; further investigation found additional points of breach. The timing does not establish a connection between the two attacks. It does put the value of a developer's laptop into perspective.
That does not make every IDE unsafe. It does make "it cannot hurt" a poor argument. An IDE with extensions is another set of software, updates, and permissions to trust. If those features no longer do useful work for you, removing them removes an unnecessary source of exposure. Simply closing the window is not the same thing, and uninstalling an already compromised extension is not a complete cleanup.
There is another reason to remove it: it forces you to think outside the familiar workflow. How do you validate what the AI produces when staring at a folder full of code is no longer the default activity?