Questions

Does hive replace tmux?

For most work, yes: sessions that keep running, splits, and a CLI to drive them. hive's sessions also come back after a reboot, scroll like any Mac app, and know about your agents. hive has no prefix key and no status line. hive-from-tmux, one of hive's agent skills, moves tmux sessions over one window at a time.

Does it work over ssh?

On another machine you run hive's server there, and use it from your Mac over Tailscale: see Remote hosts. Inside an ssh session, the hive CLI works as on any Linux machine.

Why Tailscale?

A server runs programs as you, so it has to know who connects. Tailscale tells it, without passwords or keys for you to manage. hive doesn't have accounts of its own.

Is the traffic to a remote server encrypted?

Yes. A remote server listens only on your tailnet, so everything between it and your Mac or phone goes through WireGuard, encrypted from one device to the other. Tailscale's relays, when they are used, carry only encrypted packets. By default the server lets in only you.

Two things stay true. The server runs your terminals, so whoever controls that machine can read them, as with any remote shell. And a plain TCP listener has no encryption and no identity: the server starts one only with --allow-anonymous, for tests on one machine. See Remote hosts.

Which agents does it know?

Claude Code reports its state through hooks, and Grok runs the same hooks. Codex, OpenCode and Gemini resume after a restart. See Agents.

Which systems does it run on?

The server and the CLI: macOS on Apple Silicon, and Linux on x86_64. The app: macOS 14 or later. hive for Android: Android 10 or later.

Why is the Mac app unsigned?

Signing needs an Apple Developer ID, which hive doesn't have yet. Each release has checksums and a signed record of where it was built; see Installation.

Does it send anything anywhere?

No. There is no telemetry and no account. See Privacy.