Herdr

Herdr is the runtime your coding agents live on. One Rust binary installs a background server that owns real terminal sessions, so Claude Code, Codex, opencode, Cursor, and 15+ other agent CLIs keep working after you detach, close the laptop, or reconnect over SSH. Y Combinator-backed, Apache 2.0, with 26k+ GitHub stars and 360k+ installs.

Key Features

• Always-running server: Terminals live inside a background herdr-server process; close the lid or drop the network and agents keep working, then resume sessions on restart.
• Agent-aware panes: Herdr reads every pane and marks each agent working, blocked, or idle, so you never hunt through tabs for whoever is waiting on you.
• Detach, reattach, SSH: Reconnect from any tty with the TUI, the herdr CLI, or plain SSH; clients can crash without the agents noticing.
• Agent-native API: The CLI and socket API are the same surface agents drive — they split panes, start each other, and wait until another agent is genuinely blocked.
• Runs what you already run: No wrapper, no replacement. Herdr owns the terminals of Claude Code, Codex, Cursor, opencode, Grok, and more — 19 agent CLIs detected out of the box.
• Plugins: 500+ community plugins extend workspaces, themes, and integrations.

Use Cases

Who Should Use This Tool?

• Multi-agent developers: Anyone running several coding agents at once who loses track of which one needs input.
• Remote/SSH workflows: Developers who want agents to keep working on a server or rented box between sessions.
• Terminal purists: People who want tmux-style persistence without leaving the terminal they already love.

Problems It Solves

1. Agent sessions dying with your laptop: Herdr's server owns the terminals, so closing the lid no longer kills in-flight agent work.
2. Losing track of blocked agents: Semantic state (working/blocked/idle) replaces the old "which pane was waiting on me" guesswork.
3. Juggling tabs for every agent: Spaces, tabs, and splits organize a whole herd of agents in one attachable layout.

Pricing

Herdr is free and open source. Install with curl -fsSL https://herdr.dev/install.sh | sh on macOS/Linux; Windows is in beta.

Advantages & Unique Selling Points

Compared to Competitors:

1. Runtime, not an app: tmux persists terminals but knows nothing about agents; manager apps (Emdash, Superset) quit with their window. Herdr is the layer underneath: a server that owns the terminals.
2. Agent-native state: Unlike tmux or Zellij, Herdr knows which pane is an agent and whether it's blocked, working, or done — and can wait on it.
3. No terminal replacement: Runs inside the terminal you already chose; cmux and Warp ask you to move into theirs.

What Makes It Stand Out:

• One binary, Apache 2.0, no account, no telemetry.
• Socket API lets agents orchestrate each other.
• 500+ community plugins and official agent integrations.

User Reviews

"Herdr reads every pane and marks each agent working, blocked or idle, so you don't go pane by pane looking for whoever is waiting on you." — The Contextian, YouTube review

"It is a terminal multiplexer written in Rust like tmux, except it knows about your agents... Sessions live on a background server, so you can detach, close the laptop, SSH back in later, and everything is still running." — Better Stack, YouTube review

Getting Started

Quick Start Guide

1. Install: curl -fsSL https://herdr.dev/install.sh | sh (Homebrew, Nix, and manual installs also supported).
2. Launch: Run herdr to start the server and open the TUI.
3. Open agents: Launch Claude Code, Codex, or any supported agent inside a pane.
4. Detach and return: Close the session, come back later, and resume exactly where you left off.

Integration

• Works with Claude Code, Codex, Cursor, opencode, Grok, Copilot, and more.
• SSH into remote machines; agents run remotely with full session state.
• Socket API and CLI for scripting and agent automation.

Frequently Asked Questions

Is Herdr open source?

Yes, fully open source under Apache 2.0, built in Rust by Herdr, Inc. and backed by Y Combinator.

Does it replace my agent CLI?

No. Herdr owns the terminal sessions your existing agent CLIs run in — it doesn't wrap or replace Claude Code, Codex, or any other tool.

Does it work over SSH?

Yes. The server runs on the machine, and you attach from any tty, including over SSH, without losing session state.

Alternatives

If Herdr isn't the right fit, consider these alternatives:

• Emdash: Desktop agentic dev environment isolating parallel agents in Git worktrees.
• Superset: Run an army of CLI agents in isolated worktrees with diff review and automations.
• cmux: Ghostty-based macOS terminal with vertical tabs and agent notifications.

Tips & Best Practices

1. Use spaces per project: Give each project or domain its own workspace so agents stay grouped.
2. Watch semantic state: The blocked badge means an agent is waiting on you — check it before starting new work.
3. Automate with the socket API: Let one agent spawn, prompt, and wait on others for multi-step pipelines.

Conclusion

Herdr is the persistence layer that agent multitasking was missing: a lightweight runtime that keeps coding agents working whether or not a window is attached. If you run more than one coding agent, it's the cleanest way to keep the whole herd alive and know who needs you.