Actually Cursor is starting to converge with us as we speak! You can look at their new agents mode (which is now their default for new users) as an example.
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!
On the docket! Right now the main thing we have enabled is the file system + terminals + ai agents through remote workspaces, but yes dev environments is definitely on the agenda :)
Yes. Antigravity switched to primarily be an agent management tool (the previous version of the product became Antigravity IDE). Additionally, most advanced tools automatically spawn subagents.
biggest difference is it's terminal first, and optimized for CLI agents. we don't prescribe a specific harness and instead try to work with any CLI harness you bring.
Nice. In the right track. I made something similar, but focused on local agents, but we both have issue tracking for managing multiple project and agents in parallel. It works, I think people will be surprised when they start using systems like this.
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.
yes, bad choice on my part. the origin was i was planning for it to be a superset of all your dev tools, not thinking about apache superset at all since it was a different domain
yeah i think theres a lot of ux conventions that are starting to get figured out, but we do want to be different. At least right now most dont well support remote workspace, issue tracking, or review. I bet most of the current ux patterns will look very different in a year
At first glance, it looks similar to Conductor (https://www.conductor.build/). It seems like a lot of these tools are converging on the same general ideas.
Could you share a comparison with the other tools out there?
yeah there is a lot of overlap, we are more terminal first than conductor so you can do can use any cli agent you want. We have a lot more quality of life features around the terminal like notifications, and some things similar to tmux where if you kill the app or update your sessions stay alive and running. We also recently released remote workspaces so you can setup cloud workspaces for your agents. Id say if you like the chat experience conductor is still a bit more polished, we'll get to that level of polish soon, but if you care more about the terminal and cloud and more new integrations we are shipping superset is better.
Very excited about the remote workspaces! Perfect for me as I've set up a remote Mac as a Claude terminal. I know it's only just released but the typing is quite laggy... have you considered something like mosh? I've been using this from the terminal and it makes it feel as fast as running locally.
Note that Anthropic specifically called out that usage through Conductor will be metered as "programmatic usage" in their June 15th pricing change: https://news.ycombinator.com/item?id=48126438
My colleagues are planning to get around this by using Conductor's big terminal mode. At which point I think it's basically just a git worktree manager?
I've been using this for the past few months, and I love it! It's built exactly around my workflow with many worktrees in various repos open at the same time, sometimes with different agents working side-by-side. Before Superset I just used terminal tabs but simply couldn't manage more than like 20 terminal tabs without losing track, so i coudn't scale further. Now i'm running probably 40-50 agent sessions over several repos simultaneously without any issues and losing track!
Keep up the good work guys!
one issue i'm having, maybe you know what's causing this, in long running terminal sessions I'm getting rendering issues. Like some characters show as something else. I have to resize the window to fix it, usually opening and closing the code sidebar fixes it for some time, but it never fully goes away.
I think i also have it more on the big screen (4k) than on the laptop screen.
Could you elaborate a little on what you are doing with 40-50 agents? I use Claude, I've employed sub-agents, but I still can't wrap my head around how people are using them to that extent.
ha, I didn't mean they are all running at the same time. sorry if it sounded like this.
I open a new worktree/agent session for each new feature or bugfix. Usually I start with ideation and brainstorming before doing the actual implementation. However my priorities shift daily, so I could be starting a feature on monday, but then stop during the ideation already, pick up another task on tuesday and won't go back to the initial one until a week or a month later. Since I got it in a neat worktree in superset, I still see my open branches easily, have access to its claude code session and can resume the work much easier than before having terminal tabs.
Because of this setup I currently have 17 backend sessions, 25 frontend sessions, plus several other repos open and ready, when i need to.
That's fair! We do have more paid features (a slack integration, remote workspaces etc.) but yeah we haven't found the best balance for which tier to put each in for sure.
is it terminal on steroids some kind of? so you can manage mutiple coding agents? how many coding agents you can manage in parallel that it is still comfortable to work and code changes are meaningful
yes, we surface agent states automatically so you can see what's running or needs attention across the different workspaces. there's a set of tasks where having 5-6 running in parallel is still productive for me such as running spikes and fixing small issue.
As we're investing more into integration test and self-validating for the agents we're able to increase the number without sacrificing quality.
So far we've been growing pretty healthily all things considered!
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 :)
we monetize on teams and, in the future, cloud. the bet is that teams will want to centralize their set up for this type of work, especially shared Linear, GitHub, skills, etc.
The FAQ says "Superset has a free tier. The source code is available on GitHub under Elastic License 2.0 (ELv2), so you can inspect and self-host it subject to the license terms." - what is self hosting in this context, isn't it a desktop app? Is this why it wants me to sign into something? What exactly am I signing in to?
So we also ship a cloud service along with Superset, which enables our Linear integration, Slack integration, and our multiplayer capabilities / remote workspaces.
When you sign in, you're signing into our cloud service!
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.
That's an interesting take! Basically Linux / a computer is everything you need to ship code.
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 :)
What I've started to do is to use Zulip. You can have different agents in different chats. You can upload files and you update from your mobile phone. At first I thought it was crazy but it's nice not to have 3 different AI agents running in tmux
I used Superset for quite a while until a month ago. There were some annoying issues, with freezing and terminal not being rendered how it should be. And they did repeated fixes that didn't really solve it. Since I had work to do I moved on.
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.
sorry to hear about the issue. we really messed up on the performance and balancing that with more features. looking into the imported projects did no projects show up on the sidebar for you?
will continue working bugs and hope you'll give it a try later in the future when the product's more stable :)
Fair pushback! The space will settle down eventually, it's just clear to a lot of people that there's a lot of value to be created in this space that hasn't been created yet :)
"Please don't complain about tangential annoyances—e.g. article or website formats, name collisions, or back-button breakage. They're too common to be interesting."
This uses separate git worktrees. If we have a local dev setup involving multiple docker services, is there a recommended solution for managing those envs? I didn't see.
We have a concept of setup and teardown scripts if you're interested in checking them out! Together with worktrees, you can make it pretty automatic to making copies of your repo: https://docs.superset.sh/setup-teardown-scripts
Yeah, how many agents can you people even run at once and how much does it cost you? In company we used the monthly token quota and nowadays it's basically unusable with claude opus 4.6 on high reasoning. You can basically burn through 100% usage through a single day. How does it even scale for you with N agents and which magical plans or models do you use, where tools like this are even viable?
for clarification we're less of a agent swarm tool, and more of a launch a bunch of independent agents in isolated environments and have some nice UX to manage them too. I also havent had as much luck with agent swarms or ralph loops, but i'm sure the laps will improve them with time
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?
If you try to just instruct them they might get it wrong. If the surrounding software forces them to do it, it'll always work. E.g. it can check for merge conflict markers like <<<<<<< and re-invoke codex/claude to merge again if the previous resolution failed for whatever reason (e.g. AI hallucinated, threw up, whatever).
Also you'll need a wrapper to actually detect when merge conflicts will occur and when rebasing is necessary.
Generally, the less you rely on the AI the better. Make the AI write the code, sure, but don't make the AI be the process.
I switched over to Superset from Conductor a few months ago and haven't looked back - it's really nice to be able to use the native Codex/Claude Code TUIs without any of the bloat
And headless server support! I have gotten a lot of request to let people run the agent on a separate dev box (in the cloud or some pc gathering dust in the corner). The frontend/backend are decoupled so you can run the agents on a totally separate machine than the GUI is running.
If you have any other problems/concerns/issues/suggestions/etc, reach out!
Remote dev box is brand new in the last release so let me know if it works... happy to help you set it up (I personally run w/remote dev box) shoot me an email
looks awesome! very interesting how similar these tools all look. i do think we're at an intermediate state here with the current UI where it doesn't scale as well for the future where triggering agents and review becomes much more important than the writing code part.
sooooooooooooooon! nobody on the team daily drives windows yet so until we have an engineer on the team driving windows we probably wont be able to properly support it.
Just switched from Conductor to Superset and I'm a big fan. I really didn't like the extra stuff conductor added to the UX (the text rendering always drove me nuts).
So far so good with Superset - even as a non-engineer builder.
It's nice to see people building things, but honestly I found the demo video a bit disappointing. A bit too slow, a bit too choppy, a bit hand wavy. It didn't make me grasp why I needed this in my life.
I was looking exactly for something like that. I tried installing the Linux version and it runs but:
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.
Sadly there are so few solutions, if even one, that is trying to offer a real remote agent interaction.
That's why I've build my own. No IDE per se rather than a discord bridge, with all interaction possibilities that makes sense and a way for my agent to quickly send me files or host mockups. My usecase is to build tools, reports, do research, mock up ideas and build prototypes. That's why I don't need to see the code that was written.
That still is not it for me, for now iTerm2 tabs / tmux is plenty. This, or tools like Warp, feel so heavy for me and I get overwhelmed.
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.
yeah us and similar categories of tools prescribe a lot of our specific workflow for worktrees, review, code editing where iterm + tmux is much more flexible. we have to tow the line of what's useful as a built-in feature vs being very agnostic.
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.
Sorry if I'm missing something but is there a way to create a new Claude workspace that has dangerously-skip-permissions but starts in Plan mode? There's no mode selector in the new workspace modal
Yeah I have --dangerously-skip-permissions configured as a preset for Claude in Superset, but you can add more commands too. Just go to Settings > Terminal
That overrides plan mode though, eg it seems claude --dangerously-skip-permissions and --permission-mode plan can't be used together. So I can't kick off a workspace from Superset with a prompt that starts in Plan mode but can switch to dangerous
Conductor and t3 - We don't build on top of the SDKs and don't plan to. As a terminal first we give you the flexibility to work between the different CLI agents CC, Opencode, Codex, etc. to get the latest features.
Emdash more similar but we're planning on investing more into allowing you to script and automate the app + agents more with our CLI.
there's a lot of cases that works for me with multiple, spiking or triaging issues generally take long so i usually can kick off a few different workspaces for to check back in after a while. what i don't want to do is wait around for minutes while an agent is coding or exploring.
Even with the strongest guardrails, no llm can be left unsupervised. And believing an llm can supervise itself is madness. For every agent you deploy as a “team” you will have multiplied diminishing returns.
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.
definitely understand the perspective. I feel like theres some dev work that completely requires me to be single threaded and just think hard, but there other modes where i can jump into whack-a-mole mode and try to work on a bunch of github issues in parallel, or sentry issues, or little ui bugs and the thinking goes more into reviewing plans and prs and i dont have to be as attentive to each individual session
biggest thing is flexibility on what harness you use. we don't build on the claude code /codex sdk. instead, we just let you bring whatever CLI agents you prefer and just optimize the terminal experience for using CLI agents (track running / needs attention state, set up worktrees, split pane, etc)
Ok I used it for 2 hours and I am ok paying 50 dollars for one time. It feels so light compared to my vscode usage. I don’t know if Anthropic severs are having a good time or is it superset :)
One thing I don't understand with all of these, is how they're handling different worktree infrastructure spin-up?
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.
I always feel like I want to use something like this, but then realize my NeoVim setup + tmux + Ghostty is good enough and I am not ready to learn a whole another system for modest gains.
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.
I've been using Superset as my daily driver for about 2 months and I'm loving it.
i have created a simple setup script that reserves a couple of ports per workspace/worktree.
with that, each ticket/workspace can starts it's own webserver. Allowing me to work on multiple things in parallel.
that also allows me to use Claude+Chrome MCP to e2e test each implementation in parallel.
Besides my own repo I also have a couple open source repos open. so I can quickly ask Claude questions about it instead of looking at the docs, which are usually outdated.
All in all, I usually have ~5 workspace/tickets/agents running at the same time.
And one as Q&A agents per OSS repo
yannoninator | a day ago
What happens if Cursor makes the exact same features as your product?
saddlepaddle | a day ago
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!
toddmorey | a day ago
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?
saddlepaddle | a day ago
bobchadwick | a day ago
https://superset.apache.org/
saddlepaddle | a day ago
ddxv | a day ago
xnx | a day ago
hoakiet98 | a day ago
survirtual | a day ago
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.
[OP] avipeltz | a day ago
brod_ie | a day ago
[OP] avipeltz | a day ago
xnx | a day ago
hoakiet98 | a day ago
jimmydoe | a day ago
they all look incredibly / increasingly the same?
[OP] avipeltz | a day ago
hermanschaaf | a day ago
Could you share a comparison with the other tools out there?
[OP] avipeltz | a day ago
jnovek | a day ago
hoakiet98 | a day ago
dbbk | 20 hours ago
collin128 | a day ago
walthamstow | 23 hours ago
mchusma | 20 hours ago
motoboi | 21 hours ago
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.
micro23xd | a day ago
[OP] avipeltz | a day ago
micro23xd | 20 hours ago
[OP] avipeltz | 17 hours ago
hoakiet98 | a day ago
micro23xd | 21 hours ago
hoakiet98 | 18 hours ago
kylecazar | a day ago
micro23xd | 21 hours ago
kylecazar | 4 hours ago
tdi | a day ago
saddlepaddle | a day ago
tdi | 23 hours ago
[OP] avipeltz | 17 hours ago
pplonski86 | a day ago
hoakiet98 | a day ago
As we're investing more into integration test and self-validating for the agents we're able to increase the number without sacrificing quality.
ssalka | a day ago
saddlepaddle | a day ago
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 :)
hoakiet98 | a day ago
drcongo | a day ago
saddlepaddle | a day ago
When you sign in, you're signing into our cloud service!
drcongo | a day ago
saddlepaddle | a day ago
gchamonlive | a day ago
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.
saddlepaddle | a day ago
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 :)
abirch | a day ago
tacone | 23 hours ago
desireco42 | a day ago
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.
hoakiet98 | a day ago
will continue working bugs and hope you'll give it a try later in the future when the product's more stable :)
bitwize | a day ago
saddlepaddle | a day ago
neals | a day ago
guhcampos | a day ago
https://github.com/apache/superset
wavemode | a day ago
CalRobert | a day ago
dang | a day ago
https://news.ycombinator.com/newsguidelines.html
jerrygenser | a day ago
saddlepaddle | a day ago
Ours are a bit complex but here's an example: https://github.com/superset-sh/superset/tree/main/.superset
vmsp | a day ago
Tooster | a day ago
[OP] avipeltz | 17 hours ago
mashlol | a day ago
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?
dbbk | 20 hours ago
mashlol | 19 hours ago
Also you'll need a wrapper to actually detect when merge conflicts will occur and when rebasing is necessary.
Generally, the less you rely on the AI the better. Make the AI write the code, sure, but don't make the AI be the process.
dbbk | 19 hours ago
cpan22 | a day ago
Can't wait to see what else you guys cook up!
hoakiet98 | a day ago
Definitely some exciting stuff coming with better automations, mobile and remote workspaces
[OP] avipeltz | 17 hours ago
frenchie4111 | a day ago
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
thegeomaster | a day ago
frenchie4111 | a day ago
If you have any other problems/concerns/issues/suggestions/etc, reach out!
servercobra | a day ago
frenchie4111 | 20 hours ago
hoakiet98 | 18 hours ago
frenchie4111 | 14 hours ago
We will definitely continue to iterate / try to keep up with whatever everyone thinks is the best UX
dested | a day ago
[OP] avipeltz | 17 hours ago
eikenberry | a day ago
ilja | 23 hours ago
[OP] avipeltz | 17 hours ago
collin128 | a day ago
So far so good with Superset - even as a non-engineer builder.
[OP] avipeltz | 14 hours ago
FailMore | a day ago
tacone | 23 hours ago
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.
yawnxyz | 23 hours ago
[OP] avipeltz | 17 hours ago
SeriousM | 22 hours ago
[OP] avipeltz | 17 hours ago
hmokiguess | 22 hours ago
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.
hoakiet98 | 17 hours ago
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.
dbbk | 21 hours ago
davecyen | 20 hours ago
dbbk | 20 hours ago
davecyen | 20 hours ago
straydusk | 19 hours ago
hoakiet98 | 17 hours ago
Emdash more similar but we're planning on investing more into allowing you to script and automate the app + agents more with our CLI.
Geee | 19 hours ago
hoakiet98 | 18 hours ago
ChicagoDave | 19 hours ago
dbbk | 18 hours ago
ChicagoDave | 16 hours ago
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.
dbbk | 9 hours ago
[OP] avipeltz | 16 hours ago
ChicagoDave | 10 hours ago
Constructing software requires supervision and a continuous feedback loop from a discerning human.
darweenist | 18 hours ago
hoakiet98 | 17 hours ago
bicepjai | 18 hours ago
bicepjai | 17 hours ago
sebmellen | 18 hours ago
Just FYI the first quote on your site references a date we haven’t reached yet!
hoakiet98 | 18 hours ago
roggenbuck | 15 hours ago
[OP] avipeltz | 14 hours ago
nullbio | 14 hours ago
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.
dbbk | 6 hours ago
whinvik | 8 hours ago
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.
BrandiATMuhkuh | 6 hours ago
i have created a simple setup script that reserves a couple of ports per workspace/worktree. with that, each ticket/workspace can starts it's own webserver. Allowing me to work on multiple things in parallel. that also allows me to use Claude+Chrome MCP to e2e test each implementation in parallel.
Besides my own repo I also have a couple open source repos open. so I can quickly ask Claude questions about it instead of looking at the docs, which are usually outdated.
All in all, I usually have ~5 workspace/tickets/agents running at the same time. And one as Q&A agents per OSS repo