hckrnws
Show HN: Beehive – Multi-Workspace Agent Orchestrator
by mst98
hey hn,
i built beehive for myself mostly. it has gotten to the point where my work consists in supervising oc or cc labor at tasks for multiple issues in parallel. my set up used to be zellij with a couple tabs, each tab working in a separate dir and it was a pain to manage all that. i know i could use git worktrees but they're kind of complicated, if you don't know how to use them it is easy to mess up, and i just prefer letting agents run in separate dirs with their own .git and not risk it. while i like zellij and use it inside beehive, i dont like the tabs and i forget where i am half the time.
beehive is a way for me to abstract that away. the heuristic is simple - hives are repos, so you basically have a bunch of hives which correspond to repos you work out of. each hive can have many combs. a comb is a dir with the copy of the repo you're working on. fully isolated, standalone, no shared .git. so for work or for personal stuff, i usually set up the hive, and then have a bunch of combs that i jump between supervising the agents do their thing. if you have a big repo it takes a minute to clone, and you also need gh and git because i like the niceties of like checking if the repo is there at all and stuff like that.
the app is open source, mit license. i went with tauri because i hate electron. also i have friends and coworkers who updated to macos 26 and i dont know if the whole mem leak thing for electron apps has been fixed. the app is like 9 megs which is nice too. most of it is written with cc, but i guided the aesthetics and the approach. works on mac and there is a dmg signed and notarized (i reactivated my apple dev credentials).
sharing this to get a vibe check on the idea, also maybe this is useful for you. there are many arguments, reasonable ones, you can make for worktrees vs dirs. i just know that trees are too big brain for me, and i like simple things. if you like it, pls lmk and also if you want to help (like add linux support, or like add themes, other cool things) please make a pr / open an issue.
There's a pretty popular project https://www.conductor.build that looks pretty similar, was there anything that you were missing from that one (if you were aware of it)?
Oh this is great, did not know about this but going to check it out. I like that it also has a little git thing on the right. Thank you for sharing this.
Check out https://news.ycombinator.com/shownew
There have been 100s of this project in the last month
Good for inspiration, tiring from the volume
Point taken
>There's a pretty popular project https://www.conductor.build that looks pretty similar, was there anything that you were missing from that one (if you were aware of it)?
There's probably a dozen new ones of these per week. It's the obvious idea at this point. Eventually the model providers will do it, and that's what we'll all use.
Yeah probably. Then again, opencode is not provider-specific && I prefer it to claude code (though I do use CC for personal stuff outside work because $$) and I missed their zen black or whatever the opencode $200 is.
My dream is a more indie world, so I'm glad to see you building too.
But we don't all need to share our personal, custom agent setups like we are going to be the new sliced bread. I have my own, I think it's great and better than most out there, but I'm not going to Show HN it amidst the Claw HN submissions, if ever. I generally link to interesting pieces in comments when someone asks how I implement a particular feature.
My custom agent setup is a component in a larger developer "swiss army knife" I have been building for 8ish years. Same handle on github if your are curious, project is "hof" with a rename imminent.
The agent part is built on ADK, which I believe is relatively on par with opencode, which I also see is highly regarded. The multi-workspace feature is built on Dagger and the VS Code virtualized FS and SCM interfaces. I can browse or get a diff at any turn-to-turn span, make edits that go right back in.
[dead]
Good for you?
This is not a helpful comment, please see the Guidelines linked at the bottom of the page.
> Eventually the model providers will do it, and that's what we'll all use.
Haven't they already, to varying degrees?
I have been using Superset (https://superset.sh/) and it has worked really well to automate creating & deleting worktrees, with their own terminals, and keeping everything organized. Great for running work in parallel.
It's really just a terminal emulator w/ a bunch of extra helpers to make coding agents work well. Which I really like since it doesn't try to wrap claude or codex in it's own ui or anything tricky.
It's a shame they didn't pick a name different from Apache Superset https://superset.apache.org
Love the concept! We have some similar functionality on https://usplus.ai. Where you have a hive we have an Org Chart. Can't wait to see where you take this!
This is great. I wish however there was some sandbox support, perhaps running the whole hive inside a vm for example
I can't quite tell what this is doing besides providing multiple terminal panels from a look at the front page. Can you help explain the unique workflow better?
Yeah, the idea is that you set up a repo for a project (the hive), and then once you have the hive, you can set up multiple combs (git clones, not workspaces) and work in parallel. Suppose you have like 3-5 issues / tickets you're working on - the idea is that you can do this in parallel, in isolated dirs, and jump between them in one place. I used to have to do this in tabs in zellij / sessions in tmux and remember which sessions is which issue / ticket. Also having to manually git clone everytime was annoying. So this is an abstraction to simplify this. Does that make sense?
I think it's making sense. Many of my workflows involve e.g. multiple different repos that house different parts of something (e.g. deployment automation is over here, application itself is over there, sometimes a change needs to happen to both at once). I find myself having to work serially on tickets because two different issues might touch the same repos and so I manually maintain branches and switch around between them; this adds starting friction to my work.
Ah, if in different repos that would map to different hives, so that would require switching between hives in the top left drop down. Still persistent, but not as streamlined as clones within the same repo / combs in same hive. I have been spoiled (?) by monorepos so that's why designed for this.
Conway's Law: repo structures reflect team structures and permissions mapping.
> I have been spoiled (?) by monorepos so that's why designed for this.
I too find monorepos superior at this point in time. There are essentially the same complexities both ways (polyrepo), two sides of the same coin. I have broken slightly, having "megarepos", where most code is in one place, but a few are broken out, possibly another "megarepo". The most natural split is public vs private.
Is there a conductor type tool for non-coding agents ?
Thanks this is awesome!
Crafted by Rajat
Source Code