Demo: https://www.youtube.com/watch?v=YWDHn7gUwfg
Try it: https://superset.sh/
We’re three engineers who’ve built and maintained large codebases, and we kept wanting to work on more than one thing at a time. Once CLI coding agents got good enough we found ourselves running several of them in parallel: triaging Github issues, adding a few ui features, reviewing PRs, researching a refactor, etc.
The funny part was that we and a lot of our friends had all hacked together similar scripts around git worktrees. Worktrees are a nice primitive for this because each agent can get an isolated copy of the repo, but the workflows around them can feel pretty messy, setting up/tearing down environments and managing dev servers.
We first posted here a few months ago when Superset was mostly an open-source terminal for managing git worktrees (https://news.ycombinator.com/item?id=46368739). Since then, it has changed a lot based on feedback from people using it on real codebases, plus contributions from our open-source community. The product has grown into something closer to an IDE for managing agent work across many worktrees, repos, and machines.
The biggest thing we learned is that the hard part is not just “run more agents.” It is managing all the state around them: worktrees, ports, terminal sessions, environment setup, diffs, tasks, and PRs. Once you have five or ten agents running, the bottleneck often becomes remembering what each one is doing and actual human review. We added task / issue tracking so work can move from issue → agent → diff → PR → review without losing the context all in Superset. But there's a lot more work to improve this experience over time.
We also launched Remote Workspaces, currently in beta. The idea is that you can run coding agents on remote machines instead of using all the memory and CPU on your laptop, while still managing the work from the Superset desktop app.To support Remote workspaces, we isolated the core functionality of our Electron app into a headless Hono server such that it can be deployed into any workspaces and talk to any client (such as our desktop app, mobile, web, etc) and still provide the same interface that our desktop app has.
A lot of our next work is around making agent work easier to manage when you are not sitting at your main dev machine. We’re building more functionality into the Superset CLI, improving remote workspace flows, and working on Superset Mobile (coming soon) so you can check on agents, review progress, and steer work from your phone.
We’d love more feedback on Superset, especially if you are daily driving coding agents!
For example, if I have a webapp, I want each of the worktrees to spin up its own infrastructure, and be accessible on its own unique local url, so that I can see the changes locally for each worktree, or I can have agents automate visual checks using something like agent-browser.
Currently I use docker for my infrastructure, each service running in its own container. I have a script that has a ./app worktree create worktreename. That creates a worktree as "worktreename" and spins up all of my docker infrastructure with prefixes for things like "WORKTREENAME", and I can access all my urls at worktreename.myapp.test (or just myapp.test for the main worktree).
This is working fine for now, but it'd be cool if one of these apps was compatible with this concept so I could move over to that.
The friction I have currently is obviously things like port forwarding, in app browser etc.
I keep thinking to try out cmux but haven't gotten around to it.
1. it's behind a login wall | 2. tries to download its own OpenCode instead of using the one installed on my machine
I also tried to create a new workspace. It asked what I want to do. I tried with "create user accounts", and it proceeded to create the git worktree (without asking for its name) and sent that prompt straight to OpenCode without allowing me to choose the LLM model.
I plaud the effort - I really do - but it doesn't really seems a great experience for now.
The Codex/Antigravity desktop app alternate route feels more like the direction I can buy into it, but I still feel like there's room for another novel UX not yet done.
Perhaps the biggest thing I feel like I wish existed is a "Scratchpad" mode inside a session, Claude recently came up with `/btw` and it's been really useful. Same with `!` for terminal commands within session.
Nested thinking spaces within a chat window UX is a very tricky thing to reason about.
i do agree that there's a more novel UX than the same left sidebar with worktrees and middle chat/terminal that is not yet explored.
we've been spiking on what the UI is like to program and monitor the specific steps in the workflow from triggering agent -> planning -> coding -> review -> merge.
They say "Change the harness, not your workflow". I think that captures what nobody is doing it right, I feel like there's a large effort towards over-engineering for a workflow and assuming it to be optimal when the real lever to press should be the harness.
Anyways, just an idea, wanted to throw it back your way since you're building in this space. Good luck!
It's crazy to see how we have independently landed in the same place
Good luck to your project! Excited to see how our projects differ in the future
We will definitely continue to iterate / try to keep up with whatever everyone thinks is the best UX
If you have any other problems/concerns/issues/suggestions/etc, reach out!
It is very different from current editors and the direction they are going in. In a way, it undermines the direction they are going. Current editors aim to make engineers 10x or 100x. These editors aim at a different target than the engineers. I will leave it to the imagination on who.
So far so good with Superset - even as a non-engineer builder.
Also - one issue I've seen with other tools doing worktree stuff is they don't deal with merge conflicts automatically. IMO the agents should just automatically resolve conflicts & rebase on their own, is that a thing here?
Kitty with oh-my-zsh, lazyvim and an agent. The entire thing is an ide. If I need to refactor, query data and interact with the system I just use native tools like rg+fastmod, bash, awk, jq... Either writing myself of asking an agent to do the heavy lifting.
Linux in the agent era is a breeze to operate and reason about, so the whole thing becomes a single development environment that's really light on resources and effective.
If I could provide one gentle pushback - the same way there's utility in OMZ, lazyvim etc., there may be utility in us shipping our CLI etc. - there must exist some software we can build that'll be useful to you as well :)
Emdash more similar but we're planning on investing more into allowing you to script and automate the app + agents more with our CLI.
Could you share a comparison with the other tools out there?
GUIs slow you down, in my opinion. But having the nice visual diff is something we can't really do well in TUIs, so very welcome.
So, superset, for me (been using for quite some time now) is basically to organize my agent and terminal sessions per task and project.
I can switch context much easier and can also resume working on something days later, with all my tabs nicely available and separated.
This was consuming me before, a dozen or more tabs and windows in my computer that I don't really remember to which task each belongs to.
I think one thing to remember is that the other side of us having dozens of competitors is that if the space couldn't sustain more than 1-2 parallel agent companies, a lot fewer of us would exist. We also will have a lot of time to continue creating value for our customers in the future in new ways :)
Can't wait to see what else you guys cook up!
Question on Remote Workspace: Can the remote machine port forward so I can use a browser to see / test current state of the app on the remote machine?
Constructing software requires supervision and a continuous feedback loop from a discerning human.
I’ve iterated the process hundreds of times and even with a strong specification and tests, an llm can meet the spec and pass all the tests and still build the wrong thing.
This multi-agent workflow idea is worse than web3.
But it’s a YC thing so someone will make money.
As we're investing more into integration test and self-validating for the agents we're able to increase the number without sacrificing quality.
Ours are a bit complex but here's an example: https://github.com/superset-sh/superset/tree/main/.superset
I installed Zellij on my server where most of work is happening and local machine and this works well for me. There are other issues I have now, but overall flow is fairly natural to what I am doing.
I liked that they did integrate a lot of agent workflow in Superset but my experience was that it would just take too many resources and especially with glitches, it wasn't worth it continuing. I had a period where i enjoyed working in it. It is vibe coded electron app, 2GB! is too much for this kind of app.
I just updated to their new version... it supposedly imported my projects but I can't find anything... so... I guess this is it.
will continue working bugs and hope you'll give it a try later in the future when the product's more stable :)
they all look incredibly / increasingly the same?
When you sign in, you're signing into our cloud service!
What happens if Cursor makes the exact same features as your product?
For what happens, in our heads the end goal is building a software factory where dozens to hundreds of agents are always running - something that nobody has nailed the experience for yet. Until that's a solved problem I hope we have room to grow and build!
Just FYI the first quote on your site references a date we haven’t reached yet!
