Hive RC — Remote Control

Hive can run agents on another computer while you use the CLI and multiplexer locally. The remote machine runs its own Hive daemon and agent processes; your local client displays and controls them over your existing SSH connection.

Run a one-off remote command

Use --remote <host> before a command. The host is an SSH destination, so your normal ~/.ssh/config, keys, and SSH agent apply:

sh

hive --remote me@build-box agent list
hive --remote me@build-box

The first connection may ask to install or update Hive on the remote machine. Review the prompt before accepting. Use --yes only when you intentionally want an automated install or update.

Paths passed to a remote command, such as --cwd, refer to paths on the remote machine.

Save a machine

For a machine you use often, give it a short name:

sh

hive machine add build me@build-box
hive machine status
hive --machine build agent list

The multiplexer can list saved machines and their agents in the sidebar. Select a machine or agent to view its workspace. Useful machine commands include:

sh

hive machine list
hive machine status build
hive machine reconnect build
hive machine disable build
hive machine remove build

Removing a machine forgets its local connection record; it does not stop the daemon or agents on the remote host.

Connection model

Hive uses SSH to run hive proxy remotely and bridge the connection to that host’s local daemon socket. It does not open a public network port. SSH authentication and host-key checking remain governed by your OpenSSH configuration.

Screen updates are compressed and paced for remote links. Tune the per-client data limit with remote.frame_budget_kb in the configuration. Lowering the budget reduces bandwidth and can make screen updates arrive less often.

For connection errors, first verify that ssh <host> works in a regular terminal, then see Troubleshooting.