Herdr alternative · for teams and tickets

Leave them running. Then hand them the tickets.

Herdr keeps your coding agents alive in a terminal on your laptop and reaches other machines over SSH. Cordane keeps them alive on the machines themselves — dialed out to a hub you self-host — and puts a kanban ticket in front of each one: worktree, branch, plan, review. From a browser, or a phone.

Herdr is a runtime; Cordane is a hub. The line is drawn below.
See it

Leave it running. Then leave.

The promise both tools make, on camera: a process on a machine you own, still counting through a closed tab and a hub restart. The ticket-driven part is on the agent kanban page.

A Cordane terminal still running after the browser tab was closed and the hub restarted
Watch · 1:26
Nothing loads from YouTube until you press play. Open on YouTube · all 33 walkthroughs
Same promise, different driver

Both keep agents running. They differ in what drives them.

Herdr's line is "Run them anywhere. Leave them running." Cordane agrees with every word of it. The difference is what sits in front of the agent, and how you reach the machine it runs on.

Herdr · an agent runtime

A daemon under your terminal.

  • A background server on the machine keeps panes, shells and agents alive when the client closes — inside the terminal emulator you already use.
  • You start agents by hand in a pane; Herdr detects which one is working, blocked or idle, across 22 agent CLIs.
  • Other machines join over your existing SSH access; a socket API lets agents split panes and start each other. Apache 2.0.
Cordane · a hub for tickets and machines

A board in front of the agent.

  • A worker on each machine keeps terminals and agents alive and dials out to a hub you self-host — no SSH into the box, no inbound port, no VPN.
  • A ticket starts the agent: worktree, branch, plan, implement, review, with the same working / blocked / idle / done states shown on the card.
  • The UI is a browser tab — it works on a phone, a teammate can sit in the same terminal, and a dev server gets a shareable link.
  • Free forever to self-host on the Community tier. The hub isn't open source; the deploy config is MIT.
How it works

Join a machine, drop a ticket, watch from anywhere.

No SSH config to keep in sync and nothing to open on the box. The worker is already dialed out; the board, the terminals and the agent's state all ride that one connection.

01

Join a worker

One command on a laptop, a VPS or the box under your desk. It connects outward to your hub, so a machine behind NAT or an office firewall joins the same as one in a datacenter — no port 22 reachable from anywhere.

02

Drop a ticket on the board

Start work and Cordane cuts a git worktree on a fresh branch, runs your setup hooks, and hands the ticket to the agent — Claude Code, Codex, opencode or pi — for a plan phase, then an implement phase. Or let the autonomous column pick tickets up itself.

03

Watch from anywhere

Every card shows its agent's state. Open the space to see the terminal live — from a browser, from a phone, or with a teammate typing in the same shell — then review the diff and how far behind main the branch is, in the same tab.

The same honest limit as Herdr. Sessions survive you disconnecting and the hub restarting. A worker reboot ends them — your layout and directories come back, the live processes don't. Herdr's docs say the same about a machine restart; neither tool is a substitute for a process supervisor.

Honest coverage

Which Herdr jobs Cordane covers — and which it doesn't.

Herdr's site lists what it is for. Here is the same list with Cordane's answer next to each line, including the ones where the answer is no.

The job Cordane
Agents keep running after you close the laptop or lose the connection Covered — the worker owns the processes; your browser is just a window onto them.
Seeing at a glance which agent is working, blocked, idle or done Covered — the same four states, as badges on the ticket card and in the spaces list.
All your machines in one place Covered — workers dial out to the hub; nothing to SSH into and nothing inbound to open.
Watching and typing from a phone, or sharing a live terminal with a teammate Covered — it's a web app with a real phone layout, and spaces can be shared by name.
Giving the agent its work — a ticket, a worktree, a plan-then-implement flow, a review Covered — this is the part Cordane exists for. Herdr leaves it to whatever you pair it with.
Running any of 22 agent CLIs Partly — a Cordane terminal runs anything; the ticket-driven headless runs speak Claude Code, Codex, opencode and pi today.
Agents starting and prompting other agents through a runtime API Not covered — there is no agent-facing socket API. Agents get their work from tickets, not from each other.
Living inside your existing terminal emulator, tmux-style Not covered — Cordane's terminals live in the browser tab, on purpose. There is no attach from your own terminal.
Open source under Apache 2.0 Not covered — the deploy configuration is MIT; the hub and worker are proprietary and free forever on the Community tier, which needs no license key and makes no license calls.
Beyond the runtime

What a ticket adds to a running agent.

A runtime answers "is it still going?". A hub also answers "going on what, in which branch, and is it done?" — for the whole team, not just the person whose laptop it is.

A worktree per ticket

Start work and the ticket gets its own git worktree on its own branch, its own ports and its own terminals — two tickets never step on each other, and a multi-repo project branches every repo at once.

Plan, implement, review

The agent plans before it codes, and what it hands back lands on the ticket: the diff, how far behind main the branch is, and any files it wrote for humans — a report, a screenshot — attached, not lost in scrollback.

Work that starts without you

Give a ticket a start time in your own timezone, or drop it in the autonomous column and let the board pick it up. The run happens on your worker while you're asleep, and the result is waiting on the card.

FAQ

Before you pick one — or both.

Is Cordane a Herdr alternative?
For a team that wants a board driving agents on machines it owns, reached from a browser with nothing to SSH into — yes. If what you want is a runtime inside your own terminal emulator, 22 agent CLIs detected out of the box, and an API for agents to spawn and prompt each other, Herdr is built for exactly that and Cordane isn't. They solve neighbouring problems; this page tries to draw the line rather than blur it.
Do my agents survive a reboot?
The same honest answer as Herdr's. Your disconnecting, closing the tab, or the hub restarting changes nothing — the worker owns the processes. A reboot of the worker machine ends them: the space comes back with its layout and directories, the live processes don't. If you need a process to outlive the machine, that's a service manager's job, and Cordane's apps feature will restart a dev server for you — but not an interactive agent mid-thought.
Do I need SSH access to my machines?
No. A worker dials out to the hub over one outbound connection and nothing ever connects to it. A laptop on hotel Wi-Fi, a desktop behind CGNAT, a box in an office you can't open ports on — they all join the same way, and you reach them from the hub, not from a terminal that has to know how to get to each one.
Which agent CLIs does Cordane run?
Any CLI you like, in a terminal — it's a real shell on your machine. The ticket-driven headless runs, where Cordane starts the agent with the ticket and collects the result, speak Claude Code, Codex, opencode and pi today. That is four adapters against Herdr's 22 detected agents, and we'd rather say so than pad the list.
Is Cordane open source?
Not the way Herdr is. Herdr is Apache 2.0. Cordane's deployment configuration and docs are MIT, and the hub and worker images are proprietary software that is free forever to self-host on the Community tier — one worker, one project, two users, three concurrent agent runs, every feature, no license key. A keyless hub makes no license calls; its only call to us is an update check that sends just the release channel name and turns off with CORDANE_UPDATE_CHECK=off. Paid tiers lift the caps. If Apache-licensed source is a requirement for you, that's a real difference and this page won't talk you out of it.
Can I use both?
Probably. A Cordane space is ordinary shells on your machine and Herdr is a CLI daemon, so nothing in principle stops one running inside the other. We haven't run the two together and won't claim it works until we have.
Get started

Put a board in front of your agents.

Self-hosting is free and available now — start with the quickstart, no email required. Want us to run the hub? Managed cloud starts with a 14-day free trial from the pricing page; your workers stay on your machines.

Community tier: free forever, no license key, no account with us.