VyreAI agents
Run Claude Code on your PC. Check on it from your phone.
Coding agents work for minutes or hours. Vyre gives them a terminal you can reach from any browser or your phone: see which one is busy, hear when it finishes, and step in when it needs you.
What you can do
Everything you’d do at the keyboard, from anywhere
Watch it work
Peek at the agent’s tab from a browser or your phone without touching it. The output streams live, and you get the scrollback when you open it late.
Hear when it’s done
Switch on background alerts in the web or desktop app. When a long command finishes, or the agent rings the terminal bell, in a tab you aren’t looking at, you get a notification.
Answer from your phone
The agent asked a question while you were out? Take the tab over on Android, type the answer with the compose bar and the key row, and hand it back.
Paste it a screenshot
Paste an image into the tab from the web or desktop app. Vyre saves it on the computer the agent runs on and types its path, which agents such as Claude Code read as an image.
Talk to several at once
Group a few agent tabs and send the same prompt or command to all of them, from the web or the desktop app.
Run them on a server
Install the headless host on a Linux box or as a Windows service and run agents there. They keep going when your laptop sleeps, as the user you chose, and the server only connects out.
Getting started
Start it once. It keeps going.
- 1Put Vyre where the agent runs. Vyre Desktop on your Windows PC, or the headless host on a Linux or Windows server. Sign in to the agent there as you normally would.
- 2Open a tab with the agent in it. Run the agent in any tab, or create the tab from the web or Android with a startup command such as claude, so it is already running.
- 3Walk away. The tab keeps running when you close the browser or lock your phone. Open it again from any device to see where it got to.
Recognised agents
Know which agent is in which tab
The tab list in the web and desktop apps gives these agents a tile, so you can tell them apart at a glance. Claude Code and Gemini CLI have their own; the others share the agent tile. Any other command-line agent runs just as well, under its shell’s tile.
- Claude Code
- Gemini CLI
- Codex
- Copilot CLI
- aider
- goose
- opencode
- cursor-agent
Private by design
Your agent’s output stays between your devices
Code, diffs and secrets an agent prints are encrypted on the machine it runs on, with a fresh key for each tab, and decrypted only on your own devices. Vyre relays ciphertext and never holds a key that could read it.
Read the security model →- AES-256-GCM per tab
- X25519 key wrapping
- Ciphertext-only relay
- Servers connect out only
FAQ
Agents and Vyre
Does Vyre run the agent for me?+
No. Vyre is the terminal the agent runs in. You install the agent and sign in to it on your PC or server as usual, and Vyre lets you reach that terminal from anywhere. Any command-line agent works; the ones listed above also get their own tile.
Can Vyre see what my agent writes?+
No. Everything the tab prints is encrypted on the computer that runs it, with a key for that tab, and decrypted only on your own devices. Vyre’s servers relay and store ciphertext. The agent itself still talks to its own provider, and Vyre doesn’t change that.
How do I know when an agent is done?+
Switch on “Notify background activity” in a terminal’s menu, in the web app or the desktop app. A bell, or a long command finishing, in a tab you aren’t looking at then gets you a notification: a toast and a desktop notification on the web, a Windows notification on the desktop. The tab list also shows each tab as busy, idle or exited.
What happens if the agent’s tab closes?+
If its process ends, the tab stays in the list with its last output and exit code, and you get an alert that it exited. If the computer goes offline, you get an alert for that too.
Give your agents a terminal you can reach.
Install Vyre on your PC or a server, start an agent in a tab, and check on it from any browser or your phone. Free tier, no card needed.