A self-hosted browser IDE in one command

The DIY route, honestly

The classic recipe self-hosts an IDE server: install it, keep it updated, then solve reachability, which is the real work. A port must be exposed, so you add a reverse proxy, TLS certificates, and an authentication layer in front, and from then on a service with a shell and your code is a permanent fixture of your machine’s attack surface. It is a fine recipe if you enjoy running infrastructure. It is a poor one if you just wanted your editor on another screen.

The one-command route

CodeMote inverts the reachability problem. Run this inside a project, on any machine with Node 18+ (macOS, Linux or Windows):

The host makes an outbound-only connection to a tunnel; nothing listens, no port opens, no proxy or certificate exists for you to maintain. The CLI prints a single-use pairing code. Paste it into the CodeMote web app in any browser and the workspace is live. Stop the CLI with Ctrl-C and there is nothing left running.

What the workspace is

• A real editor over the machine’s actual files: code, markdown, images. Edits save where the project lives.
• Persistent terminals: real PTYs that survive the tab closing, so the dev server you left running is still yours tomorrow.
• Git and pull requests, from status and diff to open and merge, with credentials that never leave the host.
• Coding agents as first-class chats (Claude Code today, with tool approvals and sub-agent transcripts; Codex and Cursor are coming).

Your code stays on the machine: the browser is a remote control, not a copy. That model is covered in depth in why the tunnel is outbound-only.

When DIY is still right

If what you want is specifically VS Code, with your extension collection, served from a machine you administer anyway, then code-server remains the right tool, and the comparison in CodeMote vs code-server lays out that trade honestly. If what you want is your machine, drivable from a browser, in the next sixty seconds, that is what the one command is for.