It’s fun to see projects like this that seem largely unlocked by LLM assistance, and probably wouldn’t even get started if it weren’t for them. I know the hype cycle for LLMs seems to have hit its peak already, but I am fully convinced we are still in the early days of what we’ll be doing with them in software development. The only problem I see with projects like this is that there could be so many of them. Personally I’ve built at least 5 things that I had thought would be fun, but realistically I just didn’t have the time. Now I do, and if they only scratch my personal itches, that’s okay.
This is awesome. I've been working on a similar concept based on the Escape Velocity series and love the space trader genre in general. I'm curious how you've thought about the faction organization structures as well as information flows in the sim? Are agents omniscient about prices and postings? I ended up making in-sim information a commodity that agents could choose to sell to each other, but I didn't follow through fully baking that dynamic.
Not omniscient but I do let them update the system information from one station just because it caused me issues with goal weighing since their information was so limited they almost always chose to stay within a known route versus exploring, this helped giving them more information.
But I did think, one station info per station probably also did not make sense, technology-wise, so I thought expanding it worked.
The faction system was my answer to what makes ships or planets' goals go beyond their immediate needs, and factions came naturally next. They want to make money and shift priorities to do it through incentives. Like subsidies almost.
I was hoping/thinking they (factions) could escalate to blockades, monopolies or just more supply control, skyrocket prices and events combat to hold choke points, blockades or sanctions. That way that agents FSM is focused on those goal oriented actions and ships can focus on their well being plus their factions incentives, and the crews needs handle immediate needs.
I have also been using LLMs (mostly claude code) to develop a game with Rust and Bevy, and it seems to work very well. The strictness of rust seems to really help them get the code right, and the ECS architecture seems to make it easy for them to reason about how things work (because they only need to hold a few systems and components in context at a time).
makes me wonder if ECS will ever catch on outside the gaming industry. not sure if it's really the right architecture for web SAAS. Might be interesting to rewrite some big piece of software with ECS.
Games and other specific applications call for such unidirectional data flow with such massive amounts of "independent" entities. It isn't that it'll catch on or not outside the gaming industry, it is more of a unique solution to a typical gaming problem.
So their are apps that do use similar or ECSs architecture under the hood, it is just more problem related not a solve everything solution.
right, I guess what I mean is that in the past you would pick an architecture based on your data structure/flow, performance requirements, things like that. Now there is a new metric: how LLM-friendly it is. Probably it won't be a huge factor, since a properly structured REST server would also be friendly enough, and LLMs are getting better. Still interesting to consider though, IMO.
I'm a Rust dev, but I've been worried Claude might not be able to emit Bevy (or Fyrox) sufficiently well.
I keep seeing Mr.Doob post amazing stuff in Threejs and the vibe coding scene blowing up with all kinds of generative work, and I feel really drawn to it.
Do you feel like Claude gets Bevy well enough? I know it kicks ass at Rust, but Bevy is relatively niche still.
yeah it does fine with bevy. It struggles a bit to structure the components in the best way, so that still requires some human oversight if you want it to be good. It helps a lot to emphasize that you want "idiomatic" bevy code, and ofc it also helps a lot to spend a lot of tokens on design and review (e.g. "is this the most elegant way to structure these systems/components?"). There's also this feature called Bevy Remote Protocol which allows inspecting bevy state via HTTP, which really helps agents debug things. I had claude add a little wrapper that also allows injecting mouse/keyboard events and triggering a screenshot write to disk, and now claude can drive the whole game, which helps it verify things end to end.
I can't really comment on the performance. I haven't noticed any performance issues for the things I'm doing.
Really wish they would spend a cycle or two on the automation/economy side. I know they've been really into combat lately. It's one of those games where the scale is exhilarating and then you spend hours cursing at the screen as the NPCs completely ignore all of the logistical rules you painstakingly setup.
I think it would be cool if you can host a simulation and enable online viewers. Have the simulation accessible from a website (god-mode disabled of course), and have an agent generate some sort of newspaper or news bulletin of interesting things that happen.
Last year I've been working on a space economy sim for about 3 months before realizing I made yet another spreadsheet simulator. Idea glows so bright in head!
I just started working on some similar stuff, it’s been a dream forever but I never had time for the payoff to feel worthwhile. Would love to hear more about your Claude workflow and experience with Bevy, did you consider Godot?
I did! I am more familiar with Godot but just ran into performance issues and then switched off their node scene straight to their rendering API, which made me feel like I abandoned Godot at that point. I am sure it could be optimized more but I want to worry less about chucking rendering and more scaling the economy simulation the further it went out, thus the more things to render, the harder it was for Godot.
Bevu/Rust just made it easier to brute force that scaling, since they have smaller footprints (overhead?)for each thread/process spawning.
The whole thing originated years ago to see what LLMs were capable of. Just spent a chunk of my Fable access hacking away at it. Is it intended to be 100% realistic, no. Its more art, tho I do try to lean into physics...
Very cool, did you suggest Rust/Bevy? Claude is always recommending it to me as an exemplar, it does seem nice but moving to rust would be a project. I wonder how it gets these ideas stuck in its pseudo head.
I was kind of excited but then the "made with Claude" ruined it.
I wanted to study your code but there's little point in studying LLM code for me. Same as looking at LLM art.
It’s fun to see projects like this that seem largely unlocked by LLM assistance, and probably wouldn’t even get started if it weren’t for them. I know the hype cycle for LLMs seems to have hit its peak already, but I am fully convinced we are still in the early days of what we’ll be doing with them in software development. The only problem I see with projects like this is that there could be so many of them. Personally I’ve built at least 5 things that I had thought would be fun, but realistically I just didn’t have the time. Now I do, and if they only scratch my personal itches, that’s okay.
Yup, lots of stuff I'd love to dive into that having an assistance coder makes way more approachable on nights or weekends to get things done.
And yah, ideally there should be tons more! The shift will be interesting.
[flagged]
This is awesome. I've been working on a similar concept based on the Escape Velocity series and love the space trader genre in general. I'm curious how you've thought about the faction organization structures as well as information flows in the sim? Are agents omniscient about prices and postings? I ended up making in-sim information a commodity that agents could choose to sell to each other, but I didn't follow through fully baking that dynamic.
Not omniscient but I do let them update the system information from one station just because it caused me issues with goal weighing since their information was so limited they almost always chose to stay within a known route versus exploring, this helped giving them more information.
But I did think, one station info per station probably also did not make sense, technology-wise, so I thought expanding it worked.
The faction system was my answer to what makes ships or planets' goals go beyond their immediate needs, and factions came naturally next. They want to make money and shift priorities to do it through incentives. Like subsidies almost.
I was hoping/thinking they (factions) could escalate to blockades, monopolies or just more supply control, skyrocket prices and events combat to hold choke points, blockades or sanctions. That way that agents FSM is focused on those goal oriented actions and ships can focus on their well being plus their factions incentives, and the crews needs handle immediate needs.
Does that make sense? (Terrible at proofreading)
> I've been working on a similar concept based on the Escape Velocity series and love the space trader genre in general.
I miss EV so much. I've always wondered if an MMO version of it, like EVE but with the interface/graphics of EV, would work.
I love space trading games like EV and Elite. Can I follow your progress somewhere?
I have also been using LLMs (mostly claude code) to develop a game with Rust and Bevy, and it seems to work very well. The strictness of rust seems to really help them get the code right, and the ECS architecture seems to make it easy for them to reason about how things work (because they only need to hold a few systems and components in context at a time).
makes me wonder if ECS will ever catch on outside the gaming industry. not sure if it's really the right architecture for web SAAS. Might be interesting to rewrite some big piece of software with ECS.
Games and other specific applications call for such unidirectional data flow with such massive amounts of "independent" entities. It isn't that it'll catch on or not outside the gaming industry, it is more of a unique solution to a typical gaming problem.
So their are apps that do use similar or ECSs architecture under the hood, it is just more problem related not a solve everything solution.
right, I guess what I mean is that in the past you would pick an architecture based on your data structure/flow, performance requirements, things like that. Now there is a new metric: how LLM-friendly it is. Probably it won't be a huge factor, since a properly structured REST server would also be friendly enough, and LLMs are getting better. Still interesting to consider though, IMO.
ECS is used for simulations as well, a game is just a simulation with user input and a pretty renderer.
I'm a Rust dev, but I've been worried Claude might not be able to emit Bevy (or Fyrox) sufficiently well.
I keep seeing Mr.Doob post amazing stuff in Threejs and the vibe coding scene blowing up with all kinds of generative work, and I feel really drawn to it.
Do you feel like Claude gets Bevy well enough? I know it kicks ass at Rust, but Bevy is relatively niche still.
Do you feel like Bevy is pretty performant?
yeah it does fine with bevy. It struggles a bit to structure the components in the best way, so that still requires some human oversight if you want it to be good. It helps a lot to emphasize that you want "idiomatic" bevy code, and ofc it also helps a lot to spend a lot of tokens on design and review (e.g. "is this the most elegant way to structure these systems/components?"). There's also this feature called Bevy Remote Protocol which allows inspecting bevy state via HTTP, which really helps agents debug things. I had claude add a little wrapper that also allows injecting mouse/keyboard events and triggering a screenshot write to disk, and now claude can drive the whole game, which helps it verify things end to end.
I can't really comment on the performance. I haven't noticed any performance issues for the things I'm doing.
[flagged]
ECS is basically column storage, but extended to behavior, not just data.
What is it about the space industry life that makes us developers adore it so much? :p
X4 Foundations just got a big update if you’re looking for a game built around this industry sandbox idea.
https://store.steampowered.com/app/4382310/Warp_to_Sector_On...
I came across this recently ... hints of Tradewars from the good old days.
Really wish they would spend a cycle or two on the automation/economy side. I know they've been really into combat lately. It's one of those games where the scale is exhilarating and then you spend hours cursing at the screen as the NPCs completely ignore all of the logistical rules you painstakingly setup.
Ooooh didn't know it had an update! Thanks for posting.
Neat idea!
I think it would be cool if you can host a simulation and enable online viewers. Have the simulation accessible from a website (god-mode disabled of course), and have an agent generate some sort of newspaper or news bulletin of interesting things that happen.
throws away spec doc for my own
Are we all having this idea?
Last year I've been working on a space economy sim for about 3 months before realizing I made yet another spreadsheet simulator. Idea glows so bright in head!
Hilariously I also had a very similar idea just recently.
lmao same, I was inspired by Rimworld + Eve Online + The Expanse. But I acknowledge that ideas are cheap and execution counts, and OP executed.
I just started working on some similar stuff, it’s been a dream forever but I never had time for the payoff to feel worthwhile. Would love to hear more about your Claude workflow and experience with Bevy, did you consider Godot?
I did! I am more familiar with Godot but just ran into performance issues and then switched off their node scene straight to their rendering API, which made me feel like I abandoned Godot at that point. I am sure it could be optimized more but I want to worry less about chucking rendering and more scaling the economy simulation the further it went out, thus the more things to render, the harder it was for Godot.
Bevu/Rust just made it easier to brute force that scaling, since they have smaller footprints (overhead?)for each thread/process spawning.
You might also like my AI written zero player game/sim: https://github.com/derekburgess/simcraft
The whole thing originated years ago to see what LLMs were capable of. Just spent a chunk of my Fable access hacking away at it. Is it intended to be 100% realistic, no. Its more art, tho I do try to lean into physics...
So, it's like an aquarium, only in space. You sit and watch, and sometimes maybe move rocks around, or even introduce new species.
Very cool, did you suggest Rust/Bevy? Claude is always recommending it to me as an exemplar, it does seem nice but moving to rust would be a project. I wonder how it gets these ideas stuck in its pseudo head.
Reminds me of SpaceTraders[1]
[1]: https://spacetraders.io/
Not using big_space :(
[dead]
I was kind of excited but then the "made with Claude" ruined it. I wanted to study your code but there's little point in studying LLM code for me. Same as looking at LLM art.