Using Claude Code from your phone
A terminal on a phone sounds miserable. It is not, if the terminal is one your agent lives in — a control-key row, a chat-style message box, and the session still running on your own machine.
Your agent is mid-job on the machine at your desk. You are in a queue, on a train, on the sofa. The question is not whether you could SSH in from a phone — you could, and it would be miserable. The question is whether checking on an agent is a thing you can do from a phone at all.
With TermLink it is, because of what a Claude Code session actually looks like from the outside: mostly reading, and occasionally typing one more instruction.
Open a browser tab, not an app
There is nothing to install. Open the web client in your phone's browser, sign in with Google, and every machine you have registered is waiting in a list — each card showing that session's most recent Claude message.
Tap one and you are in its live terminal. The real one, on your own machine, still running.
Reading is the main thing
Most of what you do on a phone is read. Claude Code draws a full-screen terminal UI, and TermLink relays the terminal itself rather than streaming a picture of a desktop — so the output renders as text, at a size your phone can actually show, and it survives a bad connection instead of turning into a smear of compressed pixels.
When you reconnect, the screen is repainted exactly as the agent left it. You are never looking at a stale frame.
Typing, when you have to
Two things make the typing part survivable.
A control-key row. Esc, Tab, Ctrl+C, arrow keys — the keys a terminal UI needs and a phone keyboard does not have. They sit above the keyboard as buttons. That is the difference between "I can answer Claude's menu prompt" and "I need a laptop."
A message box. The mobile compose bar is a chat input: you type your next instruction and send it. It reads like messaging a teammate, but what lands on the other end is a real shell, receiving real keystrokes.
That framing is the honest one for what you are doing on a phone. You are not writing code on a 6-inch screen. You are reading what your agent did and telling it what to do next.
Show it, don't describe it
Paste an image from your clipboard or drop a file straight onto the terminal — a screenshot of a broken layout, a stack trace, a design mock. TermLink uploads it to the host machine and types the saved path into the prompt for you, so your next message can just point Claude at the file:
> the header is broken on mobile, see <the path TermLink just typed in>
Claude reads the file off its own disk, where it is already running. Nothing about that flow requires you to be at a computer. Files up to 25 MB.
More on that in sending screenshots to Claude Code.
Turn on Auto-Yes before you leave
The one thing that ruins remote checking-in is arriving to find the agent blocked on a permission prompt it hit ninety minutes ago. Flip Auto-Yes on and TermLink confirms those prompts for you, so the run is still moving when you look.
What it is not
It is not a notification system — TermLink will not buzz your phone when Claude finishes or gets stuck. You check in; it does not check in with you. The session list is built for exactly that: one line per machine, showing the agent's latest message, so a five-second glance tells you who needs you.
Works in any modern mobile browser. Hosts run on macOS, Linux, and Windows — see getting started.