February 25, 2026
Terminal drama, now with QR
Claude Code Remote Control
Code on your PC, chat from your phone — fans cheer, veterans yell “we had this in 2003”!
TLDR: Claude Code’s Remote Control lets you use the AI on your PC from your phone while keeping data on your machine. Commenters clash over “this is just screen/tmux” versus security and API key complaints, with some praising convenience and others predicting competitors will copy.
Anthropic just rolled out Remote Control for Claude Code, a way to run the coding assistant on your own computer while controlling it from your phone or any browser. The promise: your files stay local, conversations sync across devices, and it auto-reconnects if your laptop nods off. A QR code pops up for quick mobile access and there’s an optional sandbox (a safety box for files and internet) to keep things tidy.
But the comments? Absolute fireworks. Old-school engineers pounced, declaring it “GNU screen, tmux, and mosh—but with vibes”, basically tools that have let nerds do couch coding since forever. One veteran flexed they already use a “Termux + mosh + tmux” trifecta, while another roasted, “We’ve re-invented GNU screen in the most inefficient way imaginable.” Security folks barged in yelling “container or VM” (a virtual machine—aka a sealed room) for extra isolation. Then came the paywall drama: no API keys and it requires Pro or Max, prompting one user to admit they’re surviving with… Slack threads.
Meanwhile, convenience fans loved the one-click link + QR magic and the fact your data stays on your machine. And the rivalry gossip? Commenters fully expect OpenAI to copy this any minute. The meme energy: “QR code for your shell” and “couch coding, but make it secure.”
Key Points
- •Remote Control links web/mobile clients to a Claude Code session that continues running locally with no code or data moved to the cloud.
- •It preserves full access to the local environment (filesystem, MCP servers, tools) and keeps conversations synchronized across devices with auto-reconnect.
- •Requirements include a Pro or Max plan, CLI authentication via /login, and accepting workspace trust by running claude in the project directory.
- •Sessions can be started or attached from the terminal, provide a session URL and QR code, and support --verbose and --sandbox flags.
- •Security model: local sessions make outbound HTTPS only, no inbound ports; traffic is routed through the Anthropic API over TLS via a streaming connection.