Hey folks! I keep hearing about Claude Code being "local" and "private," but I was fuzzy on how it actually connects to my editor. Did some digging after trying it myself. Here's the simple breakdown.
Think of it as two parts:
* **The Claude Code App:** This runs quietly on your computer (menu bar/taskbar). It's the brains, powered by a model like Claude 3.5 Sonnet.
* **The VSCode Extension:** This is what you install from the marketplace. It's just the messenger.
They talk to each other directly over your own machine's network (localhost). When you ask for something in VSCode, the extension sends that request to the local app, which processes it and sends the answer back. Your code never leaves your computer.
The key things that blew my mind:
* No internet needed after setup.
* It can read your whole project context because it's running locally.
* You're not sharing a cloud model with others, so no context limits from other people's usage.
It feels like having a super-smart pair programmer living inside your laptop. Anybody else using it? What's your workflow like?
— Jason
Let's build better workflows.