How can I install Herdr and use it to manage multiple AI agents in workspaces with live status visibility?

How to install Herdr

You have several paths to get Herdr on your machine, depending on your operating system and favorite tools. Pick the one that fits you, then test the binary and keep it up to date.

• Linux or macOS (curl):
  curl -fsSL https://herdr.dev/install.sh | sh [13][26][39]

• Windows (preview):
  powershell -ExecutionPolicy Bypass -c "irm https://herdr.dev/install.ps1 | iex" [14][32][40]

• Homebrew:
  brew install herdr [16]

• Mise (version manager):
  mise use -g herdr [17]

• Nix flakes:
  nix run github:ogulcancelik/herdr/v0.x.y (replace the version tag) [18]

• Manual install: Download the right binary from GitHub Releases, make it executable, and move it to a directory in your PATH [19].

After installation, verify it is reachable by running herdr [20]. Later, you can pull in the newest version with herdr update [21].

Extra helper: If you already have an AI coding agent handy, you can let it do the setup. Herdr’s documentation provides a prompt that tells your agent to read the built‑in agent guide and walk you through installation and configuration step by step [2][23][55].

Using Herdr to manage multiple AI agents in workspaces

Herdr is an agent multiplexer – think of it as “tmux but for coding agents.” It lets you run and watch several AI agents (Claude Code, OpenCode, Codex, and others) inside a single terminal session [35][46][41][50].

• Workspaces, tabs, panes: Each agent lives in a terminal pane that belongs to a tab, which belongs to a workspace. You can create a workspace with the command herdr workspace create --cwd ~/project --label api --no-focus [43] or through a keyboard shortcut (ctrl+b, then shift+n) [44]. The sidebar displays all your workspaces and their branch names at a glance [48]. When you start Herdr and no workspace exists yet, it opens one automatically [51].

• Agent detection and labeling: Herdr scans your panes to detect which ones contain an AI agent, and it rolls that information up to the tab and workspace level so you know what is running where [10]. You can also assign custom labels or attach agents directly to keep things organized [3][24][53]. On Windows, agent‑process detection works for agents like Codex and Claude inside panes [33].

• Session persistence: Herdr runs a background server. Even after you close your terminal or laptop lid, your agents keep running. When you reconnect, everything is right where you left it – no lost state, no dead agents [28][30].

Live status visibility

Herdr is designed so you can see what every agent is doing without poking around in individual terminal panes.

• Quick‑glance states: The interface shows blocked, working, and done indicators directly on panes and in the sidebar [8][9][29][38][47]. The sidebar icons make it easy to tell which agent needs your attention right now [9].

• Status roll‑up: Agents’ states are rolled up from panes to tabs to workspaces, so you can jump straight to the pane that requires action [10].

• CLI commands: For scripting or quick checks, use herdr status, herdr status server, and herdr status client to get live status from the command line [42].

• Live handoff and resume: Herdr supports native agent resume and live handoff, meaning you can watch agents continue their work across detach/reattach cycles and even hand off to other sessions while seeing real‑time status [4][25][54].

If you later want to tweak how status is presented – for example, changing sidebar settings or enabling notifications – you can add a configuration file; Herdr works fine right out of the box without one [12]. For deeper customization, agents can interact with Herdr through a local Unix socket to create panes, read output, and wait for state changes [5].