Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Herdr: Agent multiplexer that lives in your terminal

github.com

Post reply on HN
Print view View original

41–50 of 115 posts

Re: Herdr: Agent multiplexer that lives in your terminal

#41

Post by gb2d_hn »

mintflow wrote: post #35

Will people really need a bunch of different agents? Even using two codex session for different project make me feel a bit overwhelmed, or may be i just old

Same. I haven't been able to see how people let agent loops run without significant steering and produce good quality software. VS Code with one or two integrated terminals running is fine for me. Or a couple of VS Code instances if I'm working one a couple of projects concurrently. The advantage of VS Code is the code / diff visibility if you like to be hands on.

Re: Herdr: Agent multiplexer that lives in your terminal

#43

Post by frumiousirc »

boxed wrote: post #6
andrijaskontra wrote: post #2

Just use tmux no?

How would you know if a tab that isn't frontmost is waiting for input or complete or whatever?

It's not so straight-forward. Either you must add hooks the the agent to notify tmux directly or you must use an external tool that polls tmux to determine one of its panes has gone silent and then based on that, send notification to tmux.

The poling requires tmux (not screen nor dtach, as far as I could find). And, silence for N seconds is just that, the poll doesn't know if that really means waiting for input or something else. With agents (like claude) that have a throbber/spinner going while "thinking", silence is a good indicator.

Kitty terminal can be polled for current text and then see if that has changed in N seconds. This would allow not having to depend on tmux which may be preferable to some kitty users. Generally, using tmux always surfaces some annoying problems for me.

Re: Herdr: Agent multiplexer that lives in your terminal

#46

Post by azuanrb »

JSR_FDED wrote: post #18

How do people use terminal multiplexers together with vim? Ctrl+B is so hardwired in my fingers for scrolling back one screen that there's no way I'm remapping that one in vim itself. So then you have to remap that in your terminal multiplexer, while at the same time there's a bunch of people saying never change the leader key... Curious what vim users especially do about this?

I'm using tmux + zoxide+ https://github.com/joshmedeski/sesh. Then on Ghostty, I have keybinds for my workflow. Cmd+k to open/switch workspace. Each workspace is just a new tmux session.

Re: Herdr: Agent multiplexer that lives in your terminal

#47

Post by isityettime »

Once you get it set up, Emacs is a pretty damn good "agent" multiplexer as well. I use agent-shell with Projectile on Doom Emacs as my main workflow these days, and it works very well even if I have 6 projects open or whatever.

Claude and Codex are also both quite good at working with Emacs as well. Depending on your isolation/sandboxing strategy, they can either run commands against your session via emacsclient (a bit scary) or dump elisp in the REPL for you to evaluate. Both are really efficient in terms of giving you a fast feedback loop.

Re: Herdr: Agent multiplexer that lives in your terminal

#48

Post by zachncst »

Getting a lot of flack in all these comments, I really like this tool. Have been super easy to use and scale to multiple agents. I've had a ton of issues with tmux and copy and paste, this just works. I was using warp terminal, and even working on my own fork of it with it's recent open source status, but this has won me over.

Re: Herdr: Agent multiplexer that lives in your terminal

#49

Post by FergusArgyll »

f3408fh wrote: post #45
FergusArgyll wrote: post #44

How do you use multiple agents in one project? Isn't there race conditions? or is this for 2 separate projects at once?

Git worktrees

Oh. You're a genius. It's the same as 2 people working on one project. Maybe I'm dumb.

Re: Herdr: Agent multiplexer that lives in your terminal

#50

Post by justinwarner »

dagss wrote: post #29
hagen8 wrote: post #23

This is way to complex... Why don't just use some harness which manages all that and give u a good UI?

I have been actively searching for such a thing for months without finding it. Some are OS X only. Some will not work well with VMs/isolated agents. Currently using emdash, which is going in a great direction but somewhat new and buggy still.

Nimbalyst does a pretty good job. Some gaps but I can manage multiple agents across multiple projects well enough.
Post reply on HN