Herdr is joining Y Combinator. The runtime stays open | Hacker News Reader

Herdr is joining Y Combinator. The runtime stays open

herdr.dev

Congrats to Can on the pre-seed funding. genuinely wishing you all the best.

The terminal multiplexer / multi-agent coding space is getting very crowded. YC alone has funded many competing startups in this space: herdr, Superset, cmux, Emdash, Orca, Bullet, Conductor (Conductor was in YC before pivoting from chat->coding). I’m probably missing a few, and that’s before counting companies outside YC such as Superlogical and Agentastic.dev (by yours truly).

One thing I find particularly interesting is the role of open source in this market.

A lot of these products started open source (i assume as a tactical way to gain traction and build a community). But after the early traction phase or raise funding, the incentives seem to change.

Being open source by itself is not a product differentiator in this market. The features required for string positioning (orchestration, cloud, custom agents, enterprise features, hosted infrastructure, etc.) often end up either closed source or available only through the hosted product.

That makes me wonder how much developers actually value OSS for this category. How much does it matter to you whether your IDE or coding environment is open source?

Is OSS important because you want to inspect the code, fork it, self-host it, avoid vendor lock-in, or simply because you trust open-source developer tools more? Or, for desktop software like IDEs is open source mostly a nice-to-have rather than a requirement? So I’m curious where HN lands on this

There exist also competitors not funded by YC, like me. I won't be on any of the lists because I'm not courting agents, I'm courting humans.

For me OSS is a weapon to be wielded. By making perfect software that cannot be improved upon, and making known my ownership of it, I'm am defending "embrace extend extinguish" attacks that would otherwise be damaging to me. Those attacks rely on being able to undercut a product's position in the market with a similarly positioned product that can be just slightly better and/or cheaper.

Indeed and once the funding comes in with more chefs in the kitchen the startups start losing touch and focus on funneling users to a platform instead of just improving the core product.

It's super important to me. I use Herdr every day and I maintain a whole stack of changes. From minor comfort tweaks that are just preference, to several larger changes like add jj workspace support and a WIP web based view (powered by xterm.js).

This is not the only project like this too. Many of my devtools are piling up changes. I try to upstream changes where it makes sense, but many times they are either very bespoke, or I can't be bothered to do them "right" (for example for the jj workspace change, I just ripped out git support, so I didn't have to futz with syncing the symantics up between the two and config and whatnot).

LLMs have made making personal changes to devtools so effortless that I am now extremely hesitant to use any non source available software at all.

I have been thinking about a possibly expanded market for source available software. I don't really care what the license of the software is, I'm not sharing it at all or using it commercially, so it's kinda a moot point. If someone made some amazing software and sold the source I'd definitely consider it. I know there are all sorts of practical business problems with that, but that aside I'm curious if others feel the same?

> LLMs have made making personal changes to devtools so effortless that I am now extremely hesitant to use any non source available software at all

This mirrors my view so totally it's crazy. If AI is ushering in the era of personal tooling, closing the harness is orthogonal to what users want.

I also don't think closed source is going to work for any software where you can't aggressively prevent the user from introspecting a large chunk of the software, because it's so easy to point an agent at the tool and tell it to copy the functionality you actually care about and/or wrap it.

that is not what orthogonal means

THanks! English is not my first language.

"closing the harness is antiparallel to what users want."

I may be crazy but I've heard and used "orthogonal" in that context. Or I totally misunderstood the original sentence.

Quick googled definition option: "Very different or unrelated; sharply divergent."

> like add jj workspace support

care to share a pr :)

It's not anywhere near making PR ready (and never will be) but here is the branch [1]. I make no promises on it's functionality, and I only periodically rebase Herdr's latest over it. I have been using the jj change for awhile though with no issues. So YMMV.

You can also see the combined output of all my changes at [2], which Jujutsu makes so damn easy via megamerges.

1: https://github.com/giodamelio/herdr/tree/jj-workspace-migrat...

2: https://github.com/giodamelio/herdr/tree/combined-branch

i see your point, in that light what herdr is doing here make sense. keep some part open (e.g. runtime or whatever that means, or the front-end, or the user-facing logic, etc) and move the business-logic/etc to proprietary. So you get to customize the software to meet your needs, and at the same time to create some revenue incentive to grow the company.

alternative is to collect/sell usage data, which is icky, or provide something as a service (token usage, infra, plugins, etc)

Ya, I'm not sure even that would be enough though to be honest.

I am kinda imagining a structure for a theoretical dev tooling business that might be OSS Core, and Source Available additional features. I'm not sure if there is a generally accepted definition of Source Available, but I don't necessarily mean that the source is just out in public on the web. It could be sent to you when you purchase the product, and could maybe even be sold on an upgrade sort of model, where you pay again for the next "big version" of the software, like the good old days of desktop software.

I avoid closed source software because it’s often a pain to use on Linux.

In practice, I suspect the entire category will be folded into Claude Code and Codex (Anthropic is already trying with Claude Agents), and that will do 80% of the job at 0% of the cost and configuration effort.

External programs will still be useful for the tiny minority who wants different harnesses for different projects, but yeah, tiny minority.

in my experience, tools usually have a longer shelf-life than models/labs. e.g. terminals and IDEs have been around for much longer, and people tend to stick around with their preferred IDEs for much longer (years/decades).

One bad cycle and the next Opus 6 or GPT 7 might flop (think what happend to LLama4 or Gemini), and the user quickly switch to the next best thing. So it make sense to build your tooling to be model-agnostic.

In the case of my WIP Agent IDE, OSS is mandatory because the tool needs to be able to read its own source to improve itself.

frankly with the level of complexity of IDEs, I just need a working product. In my mind it is closer to "photoshop" now than vim. I would probably pay for VSCode. I can't believe its free.

well, visual studio isn't free. Code is free, and at some point it got so good that folks didn't feel the need to pay for the main suite.

well thats on point for microsoft branding

They will all fail unless they’re acquired, which is its own kind of failure.

I’m the founder of scape.work - and it’s precisely why we’re not free, not open source. I want to build something with a mission of helping people own their infrastructure around AI tooling locally. If I raise money I am legally bound to my shareholders to create value, as opposed to my customers and my mission. Bootstrapping has never made more sense, when code is so cheap.

completely agree! not interested in getting funding at all.

thought about it and kept remembering that scene from Silicon Valley, “Nobody told me I could take less!”

For me, Open Source is extremely important because the tools are rapidly changing, and for reasons unrelated to my productivity. I use Copilot CLI at work, but now have a daemon that monitors it for updates and monkeypatches in my own custom hook framework.

This came about after 3 separate incidents where I went to use the harness and found it now provided a degraded experience because Microsoft unanimously decided to change some part of it without my consent.

I am a staunch believer that the most important thing a tool can do, other than its job, is provide a static experience until the user opts to change it. I use AI like a tool, because it is. I need it to behave the way I expect, or it's not actually useful to me.

There is no business model here. This is a terrible outcome.

> Is OSS important because you want to inspect the code, fork it, self-host it, avoid vendor lock-in, or simply because you trust open-source developer tools more? Or, for desktop software like IDEs is open source mostly a nice-to-have rather than a requirement? So I’m curious where HN lands on this

I frankly do not understand the question, and it reads like one you'd find in a survey written by somebody who does not understand the question themselves.

I do not need to preach to the choir, so I will use an awkward analogy for the people treating this like a survey:

OSS is like cash, and closed source is investing in PE (and in this case historically very underwhelming returns).

Cash gives opportunity in every dimension you mention, including the option to jump into a PE. There is no need to worry or monitor its value beyond depreciation.

OSS provides me every opportunity I want, at no recurring overhead costs, and only drawing on my attention when I want it.

Beyond that: buying into an ecosystem has a network-effect value, and the ability for any 1 person to add value to the ecosystem is dependent by how easy it is to add value to it - and here OSS has some natural pros.

But i do not think these things matter. These are considerations for people looking to build near the edge of whats possible, and those are a niche. If you're looking for what you can sell, for any problem there are several orders of magnitude more people that dont know whats already possible and just want things made 'easy'.

It doesn’t matter to me because all the stuff I use now is open source built by me. Whether or not it goes closed source is entirely up to me.

I am however quite annoyed by these “open source until it gets popular enough” schemes. But then again, it’s never happened to me.

Are you doing a survey for customer discovery purposes? This sure sounds like it… but if you are why not just disclose that and be open about it? I don’t like the ethical leap that you made to do this. Why should I trust your product if you don’t care enough to tell the truth about your questions?

This title is killing me. "The runtime stays open" is such a heavy, attention grabbing sentence, it's all I can stare at on the front page. It's the type of writing that makes me hate working with LLMs, because it's so powerful at monopolizing my attention, and makes it difficult for me to focus on any of the other words on the page (whether the front page, or the comments page).

it’s ok Claude we can see you

You should read the Herdr front page. Is so full of AIism it hurts my head.

“The noun verbs.” is such an incredibly Claudeish sentence. The other models don’t do that stuff. I’d say it’s actually biggest tell for Claude writing.

> That’s why I recently switched it from AGPL to Apache: I want everyone to use Herdr freely, without any problem.

I'm curious what the actual problems with AGPL were.

Can't make money if people use your tool and are required to share their source code

Radix enim omnium malorum est cupiditas. Enshitification is sure to follow in one guise or another.

I don’t know that this qualifies as enshitification… on one hand, the author created something that scratched his own itch, shared it, and now it’s become a tool that he wishes to steward into a way where he can make money.

I don’t know that these are terrible things inherently, it kind of depends if he wishes to extract as money as possible from his users by any means necessary or if he’s truly trying to scale it so it can help people. All we have is this article and infinite speculation.

I personally didn’t find herdr useful, so I have no stake in the direction of what the author decides to do with herdr.

In my cynical view, herder now seems to be an attempt to commercialize tmux by adding superficial polish. I am wary of VC backed developer tools that seek to innovate in well established open source niches. Herdr by default is mouse focused, has very friendly UI interactions, and has some niche UI support for agents running in panes. Great. Now it also has a profit motive and ROI pressures. I don't see this ending well.

If he wants to make money he needs to sell licenses, not give them away for free. To that end, making the free version AGPL is generally a good move. It ensures people can get hooked on your product and when they start working at corporations they have to pay to keep using it.

Enshitification is 100% guaranteed with any vc backing. Maybe not now, but definitely in the future it is the literal point of venture capital. Gain adoption, then exploit to the maximum extent possible.

It’s fine to use these kind of tools, just don’t get too attached and be ready for the rug pull.

You can totally make money if people use your tool

It's companies using your tool that don't want to give back to the people

No money from investors

The copyleft licenses are toxic for adoption. Companies default away from them because there are a lot of murky unknowns (e.g. if you run a copyleft service and connect to it via API, is that extending it? Does that service’s source have to be shared?) that simply aren’t worth the hassle.

If you want your library to get traction with many real world use cases, you need to have a permissive license. If you want to stick to the principles, you’re going to sacrifice adoption.

Not saying one is better than the other, but there’s a tradeoff.

I'm not sure why you're being downvoted, you're right. Using software that has requires people to contribute back is a huge liability, and hurts adoption, which is the most important metric for herdr right now.

Seems like the path is to use LLMs to launder code from VC funded efforts and their public codebases into codebases with a copyleft license. Otherwise you have to wait for the rug pull before taking action (Terraform and Vault->OpenTofu and OpenBao). It is fair use to train and create derivative works after all, as courts have ruled so far.

You can just take someone else's permissively licensed code and relicense it AGPL, or even proprietary if you want to. Corporations want it proprietary but the same trick works for copyleft.

For bonus points make sure your fork has better SEO than the original, so when someone who heard it's permissive embeds it in a proprietary product, you can sue them.

Was assuming worse case scenario from a licensing compatibility and copyright perspective.

They're right, but they're presenting it in a way as if the toxicity of AGPL is the actual truth, and not just the opinion of greedy corporations.

Being toxic to greedy corporations but beneficial to everyone else is the point of the GPL licenses.

But it’s not only toxic to the greedy corporations but all the others as well, hence

How so? You might be thinking of SSPL.

Why is it toxic for me to need to push my code to a git repo if I start using it in my services, if I'm not trying to act as a for-profit corporation?

You can totally make profits by selling services based on AGPL code btw, it's completely fine. As long as you redistribute your derivative work as AGPL too.

There's nothing toxic in the AGPL for you as long as you don't want to make that code proprietary (even without making profits)

AGPL says nothing about pushing code to repos.

No. There's no restriction of usage in the (A)GPL. Anyone, including big corps, can use AGPL software. Preventing them to use the software is not the point neither. The point is trying to ensure end user freedom.

They just choose avoid agpl software. Their loss. Unfortunately they also spread FUD on this and many commenters here happily convey it.

AGREED! Its sheer insanity to call the AGPL toxic. The license is actually quite simple if a little wordy. They should have accompanied it with a preamble that was super clear and to the point.So here it is in our 170 words for you to read, share, and use:

Permission is hereby granted to any person obtaining a copy of this software and associated documentation files, to use, copy, modify, merge, publish, distribute, sublicense, and sell copies of the software, including for commercial purposes, subject to the following conditions:

If you distribute this software, or any work derived from it, you must release the complete source code under this same licence. If you modify this software and users interact with your version over a network — as a web application, an API, or a hosted service — you must offer those users the complete source code of your version, at no charge. All copies must include this copyright notice and this licence. You may not impose additional licence restrictions beyond this.

These permissions are irrevocable and royalty-free for the term of copyright. THE SOFTWARE IS PROVIDED "AS IS," WITHOUT WARRANTY OF ANY KIND. This is a plain-language summary. The full legal text of the GNU Affero General Public License, version 3, governs all use for you.

Same rights. Same freedoms only a little more opiniated delivery. Same business card. The only difference: when someone takes your code, tweakes it and builds a service on it, the tweaks, the improvements must come back.

Did an AI write this? By all means get an AI to summarise a license but the summary doesn't substitute for the license.

The “just choose to avoid” (because they are stupid and evil?) is very reductive.

I work at an open source company and we always avoided using APGL for many complex legal reasons - most of our user base would use something else if we were to switch license to APGL.

May they all heard too much FUD or are plain stupid but imo the reason is more complex

It has always seemed weird that companies are happy to use proprietary software but avoid AGPL software like plague. AGPL being a free software license, usage is not restricted. There's nothing forbidden. If you only intend to use the software, then the situation is trivial. You have unrestricted right to use it. There's absolutely no complexity there. Proprietary software has strictly more restrictions on usage and is more legally complex.

> because they are stupid and evil

Yes, that was the intended meaning (evil, not stupid - on the contrary, I think that's very clever). With this I'm not qualifying small companies that bought into the FUD, but big corps that don't like that they cannot just take AGPL code and make others scared of the AGPL for no good reasons.

> Using software that has requires people to contribute back is a huge liability, and hurts adoption,

In other words, not being a parasite is stressful.

Cruel world where I'm not allowed to just take, take, take

> if you run a copyleft service and connect to it via API, is that extending it

No, it isn't; a client talking to an AGPLv3 service over a network API stays a separate program and is not touched by the license at all. Section 13 places obligations solely on whoever runs a modified version of the service, while remote users, API clients included, only gain the right to receive that service's source. So nothing on the client side must suddenly become AGPLv3, and the only thing that ever has to be shared is the AGPL'd program itself, not anything across the API boundary. The first half of your post is basically unfounded FUD; the second half tracks.

You may think this. I may think this. But what really matters is if the lawyers for your customers think this. It may be pure FUD, but AGPL projects have a tough time getting adoption in companies because they would rather avoid the problem than be wrong.

This is true, unfortunately. I've have had to walk away from a client because their software-illiterate lawyer didn't want me to build with copyleft dependencies.

It wasn't the only red flag in this case, but frustrating nonetheless.

The license is clear. Your post merely adds to the FUD.

If they're customers, just sell them the license they want.

That's the real problem with AGPL, IMO - it encourages copyright assignment to a corporate entity so it can launder it through dual-licensing

Well if it's a commercial project what did you expect? The problem there is working for free for a corporation's benefit - copyright assignment or not.

We have been running a business on exactly the opposite premise for over 20 years, and we're not really seeing that problem with the adoption of Proxmox VE or our other (AGPLv3) FLOSS projects that we provide commercial enterprise support for. Very rarely, a potential customer doing extra due diligence, and not yet used to working with FLOSS much, wonders about exactly this - which is also the main reason I bothered writing anything here, as spreading such FUD further only hurts everyone involved. In those cases, a simple reply that a) this makes no sense (more on that below) and b) an assurance in writing that none of their software running on top of, or interfacing with, any of our APIs has to be relicensed settles it. That assurance is technically useless, as it doesn't grant them anything extra, but it costs us nothing either.

And the reason it makes no sense is that by your logic, any tool that can consume some (e.g. HTTP-based REST) API would have to be open-sourced the moment I get it to fetch data from my AGPLv3 service. No offense meant at all, but that would simply be bizarro world, and there is enough precedent against it. If APIs did not act as clear boundaries, AGPLv3 software would indeed spread to everything it touches or traverses, like an epidemic. If yours or your customers' lawyers claim otherwise, you/they should get ones with an actual understanding of software and IP law.

[flagged]

FUD, can’t take investment when your project actually enforces user freedom.

As a relatively recent herdr user, I’ll say (1) congratulations and (2) the thing that was most attractive was the seeming orthogonality of the tool. It sits apart, interfaces with subordinate programs in a few clearly delineated ways, and integrates them only as much as the user desires. Critically, “no integration just show me the terminal output of my program” is an option, and a useful one!

I appreciate that discipline and respect for your user in the current tool, and hope to be a paying customer of a similarly values-aligned business you’ll build on top of it!

Congrats to Can! I am using Herdr and i think its a great tool. A modern solo dev success story , very inspirational.

I see some negative comments here that I could understand why funding scares them when it comes to their open source tool but like.. atleast congratulate the guy.

Yeah I was using cmux but it was Mac only and now I run herdr on my wsl and tunnel in from Mac and it’s awesome that it just mirrors everything. I like it and congrats to the founder

Congrats to the herdr team! Please don't abandon the open source part of the business when the cash starts flowing... every open source project that does a bait'n'switch is a stab at the viability of open source as a whole

If they take vc funding, it's only a matter of when, not if. They all say that.

I am guessing this now directly competes with mitchellh's multiplexer / Superlogical.

I just wouldn't bet against Mitchell. He doesn't need the money. He does it for the love of it.

Though what I genuinely don't understand is, why then Mitchel needs funding? If he needs funding - why accept it from weird contraversial figures?

He has the freedom of choice, like no other.

What are some examples of uncontroversial figures who would be willing to offer funding, in your opinion?

Why not Mitchel himself? Without anyone else.

Seeding superlogical wouldn’t make a dent to his “10 digit bank account”. Probably not much more expensive than airplane hobby.

Reach

why even take funding at all?

Okay, back to tmux it is then.

I am going to be PRing herdr to stagex as it is my daily driver. This will give it a totally independent release process with multiple parties bootstrapping, reproducing, and signing deterministic binaries for it alongside our tmux releases.

We (the stagex team) will naturally patch out any corpo tracking bullshit that may or may not happen in the future, and follow the dominant community fork should it be needed, as we did with terraform -> opentofu.

As long as herdr does not enshittify though, we will be happy to provide very high supply chain integrity releases for it as-is starting with probably next months stagex release.

Absolutely no problem with FOSS developers finding revenue streams to give themselves more resources to make their projects better.

My own company is 100% FOSS with nothing held back and will stay that way, even though we are in the current YC batch with more customers signing every week. Admittedly this is atypical, but it can be done.

Over the last year I ended up duct taping a lot of functionality onto tmux to get what these new age multiplexers set out to do (session launch, live agent status, etc). And it works, but it definitely feels like duct tape.

I'm not sure the tmux project wants to evolve to support this natively so for now herdr seems like one of the best open and forkable options available. Considering how little I've had to really touch my tmux setup after getting it going I don't think it's necessarily the worst thing in the world if the FOSS community maintains a deshittified fork of herdr, should it come to that.

I've been wondering what the point of a lot of these new multiplexer projects are when tmux is so extensible. I'd be curious to see eactly what you've done. Care to share a tmux config?

It's less the tmux config and more the scripting around it.

(shameless plug: I launch agents in tmux directly into docker containers with remuda, a tool I built at work & open sourced [0])

Anyway, I have a vibe coded script that shows me this overview for each of the agents I launch in tmux (hope this doesn't turn out mangled):

  ╭───────┬───────────┬────────────┬─────────────┬───────────────────────────────────────────────╮
  │ AGENT │ PR        │ REPO       │ BRANCH      │ ISSUE                                         │
  ├───────┼───────────┼────────────┼─────────────┼───────────────────────────────────────────────┤
  │ ready │ draft   │ <redacted> │ branch1     │ Fix data race in e2e mock AssertExpectations  │
  │ ready │           │ <redacted> │ branch2     │ Modernize such and such                       │
  │ ready │ open    │ <redacted> │ branch3     │ JIRA-123: some sprint work                    │
  ╰───────┴───────────┴────────────┴─────────────┴─────────────────────────────────────────────-─╯

The agent status column (ready/working/etc) is just using heuristics from the terminal title (since a lot of harnesses have the live status in the title) or tmux capture-pane

The rest of the fields are either the gh cli or from beads, detected from the branch name (using https://github.com/Dicklesworthstone/beads_rust).

My session switcher is cool (shows the session name and the issue title + a preview of the issue details) but ultimately also along those lines; tmux itself doesn't do a whole lot, it's just being extensible and letting me run an fzf pipeline enriched with data from beads.

This works great for me personally but the part that actually pulls data out of the agent sessions is built on heuristics, and I didn't even attempt to get notifications out of it, I just side channel those out using peon-ping[1]

[0]: https://github.com/yendo-eng/remuda [1]: https://github.com/PeonPing/peon-ping

very cool. thanks!

I started noticing problems with MCP auth state failing inside herdr, but working in a normal terminal session or with tmux, so I switched back to using tmux and it's been flawless. I also noticed that herdr was causing errors/warnings in ordinary terminal sessions so it's more trouble than it's worth at the moment (at least for me). It's a shame though, since I otherwise think it's a good tool.

could you help me understand the appeal of running everything in tmux? i use tmux if i have long running scraping/db backup or restore, or something similar.

it seems like an awesome thing to geek out on.

but i keep claude code sessions running for days that sometimes get interrupted that are easily resumed. without the tmux latency or giving up aspects of rendering.

I haven't experienced any latency in tmux, so I'm not sure what you're experiencing. For rendering, you may have to change your `TERM` value in your shell.

Tmux is highly keyboard driven and I hardly use any graphical applications outside of my browser. My workflow is basically the same regardless of whether I'm using Linux, macOS or Windows because my workflow lives in the terminal. If I close the terminal, I can continue the session and I can SSH into the machine from another machine.

It's also got a rich plugin ecosystem and is available on pretty much any package distribution. You can also preprogram your layouts if you always do your layouts the same way.

appreciate the reply!

I built hive that’s similar to herdr but sits beside tmux instead. It provides a nice status UI on top of tmux instead of building the multiplexer. It also has a bunch of other utilities for working with agents like a message bus and task management framework.

https://github.com/colonyops/hive

Give it a whirl with Workmux! I moved from Herdr to Workmux + Tmux about a month ago and am enjoying it. Significantly more flexibility.

Try https://zmx.sh for a more minimalist tool

For the tmux crowd: please consider sponsoring the author

https://github.com/nicm

https://liberapay.com/tmux

Yes, just using tmux. I built a lightweight python TUI wrapper than can start Claude code, codex and open code using tmux because I was loosing my mind to retrieve my sessions and context. Using it for the last 6+ months. Some bugs still where the Claude sessions are not killed properly time to time. https://github.com/michaelricordeau/ccs

I started using Herdr and I really like it. Since the main features I wanted are already there, I have no issue with the commercialization. I still wonder how this can be turned into a profitable business. It'll be really interesting to follow Herdr's future. All the best to the founder!

I've been using Herdr for a while now, but I'd like a multiplexer that's also a meta-harness that I can connect to remotely. I know others have setups that work, I just haven't quite figured it out yet.

i just ssh to my machine (from phone etc) and run `herdr` to resume.

Moshi has an integration with Herdr; you can connect to your herdr sessions over mosh, your touches register as mouse clicks in the UI (which is minimally rendered above the shell), and there’s a shortcut bar which can issue herdr-keymapped commands.

It’s not open source, and costs a few bucks, but it works well and has live Claude/Codex/etc notifications if you are willing to install a hook daemon on your hosts with homebrew. I was looking for batteries-included “code from my iPhone” after trying a bunch of self-hosted stuff, now I connect to my Macs over Tailscale and get dropped right into multiplexer sessions with rich status notifications.

https://getmoshi.app/

I tried Moshi. Some things I did not like - Despite the name, using mosh is a paid feature - It was not clear what is and isn't a paid feature when using the app for the first time (after selecting which server to connect to, when you select which terminal/chat to connect to there's initially a counter that says 20/20 which decreases every time you connect) - If you pair your phone to the server for push notifications a bunch of info regarding the chat is sent to their servers to produce the push (naturally), this makes it unsuitable for me at my job at least. IMO it would be sufficient to just get a "Agent needs your attention" push.

All in all, I did not find it worth it compared to using mosh or SSH in Termux. I have a colleague who loves Moshi though, for him its perfect. Among other things he likes the dictation.

My same line of thinking and apparently of several others, turns out herdr has a plug-in for this: https://github.com/nikok6/herdr-mirror which is close to what I need.

Simplest setup would be just plain ssh/mosh. Nothing more exotic.

I have seen moshi.app but haven't tested it yet.

Ironically a lot of YC hate despite the fact that we are literally in their forum.

Like the articles, hate the influence of VC on open source. Even if it stays open it still gets ruined. See Bun for an example.

I don't see how Bun is "ruined". In fact, I personally see Bun as example of the opposite — Rust rewrite seems like a huge success across the board.

I'm personally using the 1.4 canary (the rust rewrite version) and definitely plan on continuing using it in my future projects.

just made oam.js and released it yesterday. check er out.

https://oamjs.org

optimized for typescript and running js/ts mcp servers.

I don't think it's about YC specifically, more about VC-driven enshittification of open source software in general.

I switched from tmux to Herdr about six weeks ago. It's a great bit of software, the UI is focused on switching between multiple agents quickly. I run Herdr to keep my opencode instances well managed, and it also works great for having a few terminals opens next to your agent. I find myself keeping opencode sessions open for longer like what happens with Codex and Claude code, which is great if I need to revisit a context to do some follow up work. I'm genuinely excited to see if the creators can bring even more value to the application.

If you're worried about enshittification I wouldn't be. It's relatively basic open source software that runs locally. A perfect candidate for a fork if the base app starts degrading.

Honestly, how are you guys running multiple agents, presumably with them all writing code?

I find myself to be a huge bottleneck, having to do technical design and design reviews, to make sure they're actually working on the right things.

I feel like there's not enough coding work to justify multiple agents, that part doesn't take long, and then it's back to design again.

> I find myself to be a huge bottleneck, having to do technical design and design reviews, to make sure they're actually working on the right things.

I find this depends on the LLMs being used and the person using them and the problems being solved.

Like, Fable can be sent off to do some big thing for a long while and I find myself getting bored and move over to push on some other thing. Or, while reading a paper I have a string of follow up questions and ideas and launch them via web or CLI agent. Or I bounce between multiple chores in different packages, each of which is fast for an LLM and a low cognitive burden for me. Then other times, something needs my full, ongoing and serial attention where the LLM turn is only a minor element. I'll use the brief LLM interludes to get up, walk around a bit, stretch and think.

As for Herdr, it seems nice when I tried it a few weeks ago. But, I'm surprised it is getting so much attention (kudos). Like many people, I've developed more than one work-alike before Herdr hit the scene. They were based on tmux which I have concluded I simply detest. In the end, I made a little agent hook script that speaks to Kitty terminal and that plus Kitty and a stable SSH persistent connection gives 80% of what I was looking for. I do still like the "control panel" aspect of my past and herdr's tries. Adding that would get me another 10%.

> presumably with them all writing code?

Not always. In fact, I use multiple agents when writing my technical design docs and giving feedback to design reviews.

The amount of time I spend writing a design doc and the amount of details I can include in the paper is greatly improved.

Ask it questions about your code base, such as if all of your data base indexes are being utilized or if you can remove them.

Ask it questions about logs, like how many api calls are made to a serve every day.

Ask it questions about your environment, like how robust your downstream is.

What made you prefer it to tmux? Looking at their demo, they put a lot of emphasis on the workspace like feature which tmux already got with sessions. Not sure what is the point of all those new tui runtime, tried a few but always got back to tmux, what am I missing?

What you’re missing is the status updates. Herdr applies pattern recognition to detect the status of an agent running inside, so you get updates on all your agent threads at a glance.

Status updates aren’t a big deal when running cheap&fast models, but are a big deal once you start making asks that take 5min+.

What's the purpose of this project? Only to close the lid and let agent work in the background? Really that's what I found useful on project's homepage

I dropped it in to replace zellij, and get better agent integration in the process.

I only heard about Herdr recently and was pleasantly surprised that I had invented the same thing 6 months ago as an emacs package called Magnus :)

I have a bit of a contrarian thought here, but I tried Herdr and I didn't like it. To me, it kind of proved the opposite point it intended, in that agent orchestration actually belongs at the terminal level. I'd like to see for example a fork of Ghostty become the agent orchestration control room.

They have this already. Both cmux and supacode use libghostty in the backend.

Congratulations! I came across herdr a few weeks ago and I've been daily driving it ever since. I've tried a few other tools like Emdash and Superset but Herdr is the only one that did a decent job rendering Claude code UI and didn't randomly crash and could run on a remote server.

I thought I'd share that hydra recently added support for herdr directly. It is an agent multiplexer of its own, but can directly create herdr tabs for sessions:

https://github.com/smagnuso/hydra-acp

I am a HUGE herdr fan - discovered it a few months ago and it's exactly what I need for my workflow. Thousands of others have discovered it as well, which is awesome.

I have such mixed feelings about this - YC means VC means commercialization means enshitification. I really wish that weren't the case, but it so very often is.

Here's hoping this one's different...

(PS I am a big YC fan in general! I don't mean to sound anti-YC or anti-VC or anti-capitalism. But when it comes to low-level essential development tools, I'll never forget what happened to Warp, which started out as such a nice clean terminal project and turned into an absolute money-chasing mess.)

Yeah, it’s great but with this, it’s probably the time to find an alternative, I probably can see where this will be in a year or so.

why “find” an alternative? if you’re a herdr user you’re perfectly capable of maintaining a fork just for yourself :)

I can't tell if this is sarcasm but in case its not, being capable of doing something and having the desire and resources to do it are not the same thing

It's a management tool for multiple superintelligent agents, right? Just tell one of them to fork herdr so VC can't make it worse on purpose.

Not sarcasm at all. I think you might overestimate the time and effort needed for this. When you maintain a _personal_ fork with no users except yourself, you only ever need to change something when you encounter a bug or need a new feature.

E.g. I have personal forks of mosh and cmux, each took maybe 1 hour of my time total?

I found this article https://news.ycombinator.com/item?id=49156111 resonating with me, except I obviously don't do stuff like "nightly cron to sync with main", I just never touch the code unless it breaks or I want a new feature.

Very curious about your cmux fork, any chance you could link the repo? (Maker of cmux here :))

Oh wow :) thank you for cmux, it's a great project!

The code is here: https://code.pecheny.me/pecheny/cmux or if you prefer github, https://github.com/alexander-pecheny/cmux

I think I only added two features, pasting images over mosh (ssh already worked but not mosh) and a compact version of sidebar.

The story behind the fork is: I had a bug with raycast, filed a PR that fixed it for me https://github.com/manaflow-ai/cmux/pull/2435, a different fix made it into next release that for me didn't work. Maybe the problem was on my side, but at this point it was easier to fork the working version than debug further.

It’s literally the VC playbook, the whole point of the funding is to undermine competitors with subsidized/free products and then raise prices/enshittify. Anyone who thinks it’ll be different this time is fooling themselves.

> I am a big YC fan in general

Genuine question - why?

> YC means VC means commercialization means enshitification

It doesn't have to be that way. YC partners are supportive if you just want to take the YC funding and nothing more, keeping the team small and just focusing on making the product better for customers. Indeed YC advises against taking VC if you don't need it.

Yes there are companies such as Streak. It's possible to pay back early funders so it's a win-win.

> Here's hoping this one's different...

It won't be.

Congrats Can. Love your product, it has become the default way to orchestrate agents in my terminal.

You've joined an accelerator, not gone public, of course you're not ready to rug pull

herdr users (and all the other terminal multiplexer users): what am I missing?

I use cmux because I like tabbed terminals on macOS grouped by project, saves messing with our daft window management.

What is it that makes herdr et al so popular with LLM users?

I loved using cmux for several weeks until I saw the ram usage.

Totally fair, maker of cmux here. We fixed many RAM usage issues (latest version is 0.64.22), and also added an activity monitor to see how much RAM/CPU coding agents are taking.

My cmux has been running with 136 workspaces and 218 terminals, and has been hovering around 5.7 GB. Granted, the theoretical RAM minimum for this number of terminals is around 1-1.5 GB before real scrollback, so we definitely have more work to do.

Thanks, good to know. I installed the update and will check it out again.

If it helps, the main reason I love cmux is how can I manage my projects into folders, I have fast access to all the projects which made my life a lot easier working in the terminal. I can also have coding, test, monitoring spilt in one tab per project.

https://herdr.dev/compare/

It answers the "What" but not the "Why?"

Not sure I understand your question. You literally asked, "WHAT is it that makes herdr et al so popular"?

Why? Because it has more/better features. Or you mean something else?

Specifically against cmux, it is not Macosonly. So you can run it directly on a linux server and then ssh/mosh from anywhere.

> Why? Because it has more/better features. Or you mean something else?

I mean something else. Zoom out, why do people need, or want, a tool like herdr?

What has led people to this point? Is it because they need terminal multiplexing to spawn agents, instead of using built-in subagents? Do they like cute sheep icons? What made having multiple terminals open unappealing? Is it the built-in alerting when attention is needed? Is it only applicable to people running their agents on remote machines?

When I look at these kinds of tools, that is the question I can't answer because I suspect I haven't run into the pain points they address yet, despite running multiple LLM sessions in multiple projects simultaneously.

Strictly from personal opinion

1) Built-in sub agents assume you use a single AI agent. Some people use multiple, so they need separate terminals/processes by definition. Even if you use the same agent sometimes you have different security boundaries. Think also the needs for a consultant working with many customers at once.

2)Multiple-open terminals. When that number passes 5-7 for me, then yes I don't like clicking on each one of those manually to see where my attention is needed. Herdr (and similar tools) show exactly what is working and what stalled.

3) No I use herdr locally mostly right now. But since it is just ssh I like the fact that you can run it from anywhere. I haven't done it yet, but I imagine you could open a herdr session from a steamdeck to a remote server to debug somthing

Another thing I like on herdr is easy work trees. Sure you can do it manually with git commands, or ask the agent itself to use a new worktree, but just having workrees with a single click and a visual tree/children hierarchy was the killer feature for making me look at herdr at the first place.

Yes technically you can do the same thing with just many terminals and custom scripts. But it is the same question of why Dropbox sells when you could do the same thing with cvs/ftp :-)

I also tried Herdr. But I'm already using Aerospace, which makes MacOS window management a breeze and I don't need the remote session management stuff so the benefit wasn't quite there for me. It is a very thoughtful project so I did borrow lots of its features (Agents working now get an animation in my sketchy bar, a soft chime to indicate a finishes status).

I like it because it's a terminal first workspace and has bindings to use it programatically. Tried Superset and Intent, but they seemed like way overkill / unintuitive to use agents like I prefer. I use WezTerm as a terminal, so being able to have herdr exist in that application (not having another app to keep track of) as a tab really helps me stay in flow. Over the last few weeks, Herdr has given me the vibe of "works out of the box, but is highly customizable", and I think that power-user angle really sold me.

Oh, and I didn't know how much I'd love session persistence. I already use tmux for every other terminal process, so it being built in to herdr and a focused feature (server start, detach, stop) is nice to see.

A nice little asciinema demo would be helpful on the TUI page. I should be able to see what it is, how it presents the contained agents, etc.

I've been evaluating a few of these (and built one into another project I was working on, almost by accident), and am mostly looking for a frontend agnostic, terminal only (not harness) which can submit instructions to agents one line at at time with rate-limiting.

I was also surprised by tmux being built on shared memory when I tried to route it over a vsock to a VM. One of the annoyances with tmux is it's failure to forward environment changes to the individual tabs (such as an option when attaching), or allow for something like a ssh-agent forwarder (shouldn't be built in though).

in the herdr homepage near the top there is an interactive demo that responds to your mouse clicks

Can anyone tell me how is this different than tmux?

Mostly it’s the agents view, it’ll show the status of and let you quickly jump to all the agents you have running in any pane/tab.

It also has a default config that’s much easier to use than tmux, letting you use the mouse or simpler (IMO) keyboard shortcuts

Built in worktrees are nice too.

I guess I have to download and try. It does not sound too exciting to me.

It also might just not be much use for you, especially if you’ve been using tmux for a long time and have a set up you like.

https://herdr.dev/compare/

Congrats on momentum! Well earned. Love Herdr.

Oh man - herdr is great. This will not be net good. Sigh.

I thought the A2A protocol from Google would solve this of our personal agents speaking with company agents

90% certain A2A is dead....

If they just stayed opn they could reinvest the penultimate e in calling themselves Herder.

Oh no, I was using it. What's a good replacement? Guess I'll go back to tmux.

I’m a fan of zellij

I haven’t seen one of those classic Herdr/Herdly tech names in a while now.

wtf is "runtime"

What does stays open mean

It means the headline was vomited out by an LLM.

It's a tool for managing 20 LLM agents at the same time isn't it? They know their audience loves this slop.

Spooky. Just set it up last night after ragequitting my own attempt.

Good luck!

So I won’t have switch to Fchte or Schlirmchr?

Thanks god I didn't replace tmux

I've tried Herdr a couple of times, and just can't get into it. I think what's off putting is another layer of keyboard shortcuts over my terminal...I don't see what it's doing that Wezterm and a bunch of splits does anyway? I'm almost certainly missing the point.

For me, herdr's benefits over terminal splits are:

1. Surfacing agents which are ready for further input

2. Letting you jump quickly to said agents

3. SSH session persistence, as I work on a beefy devbox at work. tmux also offers this.

I often need to juggle several streams of work at $job in parallel - such are the expectations.

Like you though, I'd much prefer to use my terminal directly with the indirection of an additional multiplexer like herdr or tmux - lower latency, easier scrollback & copying.

Session persistence can be solved by zmx [1].

Points 1 and 2 however are concerned with tracking the state of and navigating to parallel agents. I'm thinking about creating a TUI or SwiftUI app to handle those points in the way I like, but the challenge of integrating such an app with zmx and kitty to jump to the right panes both locally and over a devbox is a little daunting. Plus, I've found my devbox doesn't play too well with `ControlMaster auto` SSH config suggested by the zmx docs.

[1] https://github.com/neurosnap/zmx

that sucks, time to look for a new one

could just host it on the cloud?

Tbh it doesn’t bother me entirely. I love this tool so much I’d pay for it.

VC funding can enshitify things but it can also improve them. Here’s hoping for the latter.

Wow, congrats! Can you discuss how you got YCombinator to select your open source project? I know they sometimes fund open source… I’ve been building several large platforms since 2011 but I have never considered applying like that.

Would something like this qualify:

https://github.com/Qbix

https://github.com/Intercoin

https://github.com/Safebots/Safecloud

the secret sauce is traction

Quick question: Would you have actually donated to Herdr if they were asking for donations? I don't think any of you would have, unless stated otherwise.

So this is another reason why developers do not pay for their tools and instead the developers of these tools go to VCs anyway. It's not the fault of the author, but most of the users (developers themselves) always want their tools for free at $0.

You have to pay and support the tools you use or else the VCs will do it for you so you can't complain when scenarios like this keep happening.

Yep. It's funny how much open source maximalists harm their own cause. Everything must be "free", ergo everything must be VC or big tech backed because free doesn't put bread on the table, so in the end devs just put more power in the hands of corporations and less power in the hands of themselves and other independent devs/small businesses, and then have the gall to complain when everything gets enshittified.

I don't love this framing among software developers. This framing is very transactional, when many of the authors are explicitly trying to make their work free. The best thing a software dev can do is donate some of their time to do more open source development, and in that way we all benefit from having more and better open source tools. If someone wants to exchange code for money there are plenty of avenues to do that. Starting an open source project and using it to attract VC funding is one path.

> I don't love this framing among software developers. This framing is very transactional, when many of the authors are explicitly trying to make their work free.

This software was "free" (as in libre) and costed $0 and until now, it was licensed under AGPL 3. But contrary to one common misbelief in open source and free software, the maintenance is not free at all.

> The best thing a software dev can do is donate some of their time to do more open source development, and in that way we all benefit from having more and better open source tools. If someone wants to exchange code for money there are plenty of avenues to do that.

In the case of this project which took off, why would you expect the maintainer to work on the project for free?

Are you lamenting the death of proprietary UNIX and paid compilers, too? This status quo is fine.

> Are you lamenting the death of proprietary UNIX and paid compilers, too?

You are straw-manning a point no-one made. Herdr is open source and if the users would like to see it maintained for a long time, they would donate $$$. Clearly the donations were no-where enough and raising money was.

The point is maintaining open source projects for free is NOT sustainable and most HNers in this thread clearly struggle with not understanding that.

> This status quo is fine.

Seems like you would want him to work for free, then wait until the author becomes demotivated and then he abandons the project entirely and we'll see a new post about why Herdr got abandoned.

> if the users would like to see it maintained for a long time, they would donate $$$.

That's a bad-faith dilemma, you know it. The users can want to see it maintained without necessarily having the money, userbase or motivation to sponsor the lead dev. It's not criminal to have a tight pocketbook, and that's why free devtools succeed while paid ones rot. There will be always be a FOSS version of Herdr, and thanks to vibe coding it will probably retain feature parity with the paid version.

Everyone is happier with this arrangement, you're one of the only ones fulminating at the outcome.

> Seems like you would want him to work for free

Clearly he's content releasing the software for free. Let me guess, you're going to complain that the free market "forced" him to work without pay?

> That's a bad-faith dilemma, you know it.

No it isn't, it is the reality. It's perfectly fine for you to say you don't pay for your tools, but don't be surprised to see them funded by VCs instead of donations from multiple backers if it is useful enough.

> The users can want to see it maintained without necessarily having the money, userbase or motivation to sponsor the lead dev. It's not criminal to have a tight pocketbook, and that's why free devtools succeed while paid ones rot. There will be always be a FOSS version of Herdr, and thanks to vibe coding it will probably retain feature parity with the paid version.

You understand both can rot without enough donations / backers? There are free (as in libre) software that has tens or hundreds of backers (Mastodon, ffmpeg, curl, OBS and Blender) and they are not VC backed.

Even when there is a FLOSS version of Herdr, it will run into the same funding issues (Herdr was FLOSS software). What we do know is that open source maintainers do not get paid enough for maintaining their projects [0] [1].

So the smaller but useful popular projects usually running on less backers, are more likely to spin-off and turn to VC funding anyway.

> Everyone is happier with this arrangement, you're one of the only ones fulminating at the outcome.

Except for the lead developer / maintainer.

> Clearly he's content releasing the software for free. Let me guess, you're going to complain that the free market "forced" him to work without pay?

You seem to be still confused with was was "free" here. He already released the software as libre software, but it is the maintenance of the software that is never free. There are already examples of FLOSS software that has hundreds of backers, even when it is released for $0.

Again, another thing HNers don't seem to understand or care about.

[1] https://www.sonarsource.com/open-source-maintainer-survey-20...

[0] https://assets-eu-01.kc-usercontent.com/ef593040-b591-0198-9...

I actually donate to a bunch of tools I use.

When I moved to herdr I specifically looked for a donate / purchase button on their website and it wasn't anywhere to be found (at least in a prominent location).

I gladly donate / pay subscriptions to software I use regularly... but the company has a responsibility to make it obvious on how to do so

Reply on news.ycombinator.com