Made for your agents.
Remade by them.

A native home for Claude Code, Codex and every coding agent. Every pixel is a file your agent can change. Close it, they keep working.

$ curl -fsSL https://releases.herdr.dev/app/install.sh | sh

Paste it in a terminal.

All downloads and what happens next →

Same agents. Ask for another look:
❯

A new kind of agent workspace. Native and terminal first, with nothing between you and your agents. Every part of it is a plain folder your agent can rewrite, and every agent keeps running when the window is gone.

Your agents run agents.
And hear back.

An agent hands work to helpers and keeps going. When they finish, their reports land in its input. No copy-paste between windows: herdr-core carries the messages.

It starts helpershc run --bg codex "review my diff. Done: hc send t1"Any agent can start another: Claude, Codex, a test run. It says how to report back.
It steers themhc send t3 "take another look"While a helper works, your agent can redirect it or answer its question.
They report back to it<hc from="t3 reviewer">Reports, exits and "needs you" land in your agent's input. You only hear the result.

##Know who needs you

Every agent tells Herdr what it's doing: working, done, or waiting on you. Herdr shows you which one needs you, in the sidebar and on its tab.

The rest stays quiet, so you can stop checking. A space for each repo and a tab for each piece of work keep the agents with what they're doing.

##Close it, they keep working

Your agents don't live in the window. They live in herdr-core, a small server on your machine. Quit the app, update it, crash it: every terminal is where you left it.

 

More about herdr-core →

##Every machine, one view

Your laptop, a server, a VM, a cloud sandbox. Add any machine you can reach over SSH, and its spaces and agents join the same sidebar, under its name. Start an agent there the same way you would here.

# or ask your agent: "add my server vps"
hc machine add vps

Its agents keep running there while your laptop sleeps or the app is closed.

##Ask for anything

Everything you see is a plain file in ~/.config/herdr-app. Its AGENTS.md tells your agent what it can change. No plugin system, no fork to maintain. You ask, it edits.

~/.config/herdr-app/
├── AGENTS.md        what your agent can change
├── settings.json    fonts, theme, sounds
├── keys.json
├── layouts/
│   ├── main.json
│   └── board.json
├── mods/
│   ├── agents/main.js
│   └── tabs/main.js
└── themes/
❯ put a timer on each tab: how long its agent has been working

⏺ Update(mods/tabs/main.js)
  ⎿  Updated mods/tabs/main.js with 6 additions and 1 removal

⏺ Done. Each tab shows how long its agent has been at it.

##Any agent

Herdr runs real terminals, so every agent works the way it does today. Any agent that speaks OSC 7501, the open Program Status Protocol, joins the herd with its status: working, done, needs you.

# one line, and your own script is an agent too
printf '\e]7501;state=blocked:app=my-agent\e\\'
Claude Code Codex Pi OpenCode Gemini Copilot Cursor Amp

##Layouts

None of these ship in Herdr. They're examples of what you can ask for. Copy one line, and your agent adds it next to yours.

See all layouts →

###Share yours

  1. Ask your agent to write it up. One prompt makes a RECIPE.md: the brief another agent could build your layout from.
  2. Push it with a picture. A public repo with the layout, the recipe and a screenshot, tagged herdr-layout.
  3. It gets a page here. People copy one line and their agent does the rest.

##Open source

The app, herdr-core and Herdr TUI are open source under Apache 2.0. Read it, change it, build your own client on the core.

herdrdev/herdr-app →   herdrdev/herdr-core →   herdrdev/herdr →

##Questions

What is Herdr?

A native app where your coding agents live. They run in herdr-core, so they keep going when the window closes. Everything you see in the app is a plain file your agent can change.

How is it different from Herdr TUI?

Herdr TUI is the original Herdr: a terminal app with its own server. This is a new, separate program built on herdr-core. They don't share agents, so what runs in one doesn't show up in the other. Herdr TUI keeps working, and everything for it is at herdr.dev/tui.

Which agents does it work with?

Anything that runs in a terminal. To show an agent's status (working, done, needs you), Herdr reads OSC 7501, an open protocol programs use to report it. An agent that doesn't report it still runs, just without a status.

What happens when I quit, update or reboot?

Quitting or updating the app changes nothing: every agent keeps running in herdr-core. If the machine reboots or herdr-core stops, the programs stop too. Their panes come back with a shell in the same folder, but Herdr doesn't restart an agent or resume its work in progress. Start it again and carry on.

Which systems does it run on?

macOS (Apple silicon), Linux (x86_64 and arm64) and Windows (x86_64). Other machines join over SSH, whatever they run.

Is it free?

Yes. The app, herdr-core and Herdr TUI are all open source.

Made for your agents.
Remade by them.

$ curl -fsSL https://releases.herdr.dev/app/install.sh | sh

Paste it in a terminal.

All downloads and what happens next →