Connected machines

Make your computers part of the same AI team.

Connect Termdock installations, mount shared workspaces and run agents on the machine that owns the project. Read output, send tasks and take over a terminal without moving the whole project to your laptop.

Workflow map · not a product screenshot
  1. 01Laptop · coordinate
  2. 02Peer computer · run agents
  3. 03Shared terminals · hand off work

Before you start

  • Install Termdock on both computers. Start on a trusted network where the devices can reach each other.
  • Install the required CLI agents and sign in on the computer that will run them.
  • Choose which project workspaces to share. Pairing does not copy files or transfer AI subscriptions.
01

Pair the computers

Open Peer Sessions on the host, enable sharing, then choose that device from the other Termdock installation. Compare the verification codes and confirm the pairing on both sides.

02

Mount the project, not a second copy

Choose a shared workspace from the workspace picker. A terminal opened in that workspace runs on the other computer, with its files, installed tools and credentials.

03

Give each agent a clear part of the job

Run the required agents in their project terminals. Termdock’s CLI and Terminal API can create terminals, send input and read output through the peer connection. Have the coordinating agent pass a bounded task and inspect the response; do not assume separate conversations share context.

04

Disconnect and come back to the work

Termdock 1.24 keeps peer terminals in the host’s background process, so the host app can quit, crash or update without ending those terminals. Reconnect to the existing session. Host takeover and handback keep one desktop writer in control at a time.

Know the limits

  • Keep the host computer running and signed in. Background terminals do not survive a shutdown or sign-out.
  • Peer connections and SSH are different paths. SSH workspaces are supported, but SSH terminals still end with the app in 1.24.
  • Terminal control over a peer does not imply every local automation or structured agent feature is available remotely. The safe dispatch queue currently auto-sends for supported local macOS Claude/Codex sessions.
  • Share workspaces deliberately. Each agent still needs file ownership or a separate worktree when edits could conflict.

What you should end up with

Your agents run on the computers you choose, while their terminals remain reachable for instructions, progress and handoffs.

Documentation & sources

Start with the project in front of you

Give your next task a workspace.

Download for free. Choose your installer, open a project, and run the AI CLIs you already have. Model usage and subscriptions are billed separately by your AI providers.

Latest release · Free download

Release notes & all installers

macOS: open the DMG and drag to Applications. Windows: run the EXE installer. No Linux installer is currently available.

Installation help & troubleshooting