> It is made to sit in a split next to your AI agents. They can send you short messages that show up in the game, so you can play while they work.
> Agent messages
> Your AI agents can message you inside the game. A message costs the agent only the words in it, and nothing prints on success. Messages wait while the game is closed.
It's kinda amazing how the ancient "local mail" technology keeps getting reinvented, you know, and reintegrated inside the gameplay? NetHack had its its "mail demon" since... (googles) version 1.0.2, in April 1985 — the creature would spawn in the game, deliver you a scroll of mail, and vanish. And apparently, it's still in use nowadays? "On public servers such as nethack.alt.org, a user may observe other users' games and communicate with them by sending mail."
wow, i didn't know about NetHack, that's pretty damn cool!
and yeah, genuinely insane how we came full-circle lol... on terminactory it uses claude code hooks, with an optional "tfac msg" command (tried to minimize token usage, hence the hook)
but yeah, the netHack thing is pretty cool, guess you learn something new every day? xD
that's a good point, actually. the features that differentiate terminactory from.. say, factorio, is the agent telegrams, the lore behind the game (optional, it's a factory game, with a lot to discover story-wise, in-game), and some other QOL features ("blueprint" that let you copy paste sections, with mirroring and rotating, unlockable on the early game; selecting a large area and deleting it, and a few more)
of course, it's not objectively better than any game, but there are a few separating qualities from the average popular factory game xD
There are, but I was trying to make the game for a more general audience. I'd be willing to add another game model with more realism, and some more processes too, if you'd be interested.
I just don't see much appeal for such games.. If it's liked substantially by a few people, I'll add it as a gamemode, otherwise it feels like catering to a non-existent crowd :/
but, now that you mention it, I'll look into adding it soon, thanks!
I’m currently vibe-coding a game for myself inspired by prison architect + workers & resources - it’s about a fictional 1940s-1980s Soviet plutonium production complex where you have to manage… workers, resources, political goodwill, quotas, and accidents. Modelled are Alpha, Beta, gamma and Neutron radiation, effects on workers like in rimworld.
Will be gpl 3.0 and I plan to release a playable 0.1 on GitHub. I care mostly about historical accuracy of radiation incidents and factory issues at the intersection of nuclear engineering & political issues akin to Paper’s please.
I’m afraid to publish it because I despise the amount of AI slop out there. I just want to tickle my need for a historical plutonium production management simulator.
well, that is an INSANE amount of praise, thank you so much! but yes, the ASCII art was inspired by Dwarf Fortress, Cataclysm, and a few other games :D
That was CP437 squeezed into a terminal emulator that ran sixteen or so OpenGL calls per character. I spent a couple months building a shim library to optimise it, before Toady let me fix the game. :p
Real terminal output ought to be faster, but don’t sleep on Kitty.
If it runs fast and low resource, it could be a nice testbed for agent architectures. I'm thinking old-school planners, genetic programming, ML techniques, etc.
Especially integrating planners or logic-based approaches with ML models. The old architectures for StarCraft were great at building and micro-managing units but terrible at assessing or adapting to enemy strategy. Hybrid architectures can address that.
Games like yours, esp different styles or requirements, can be used to prove how well the techniques generalize.
that's a good idea, I could definitely see this and games like this as agent benchmarks, like an ARC-AGI but it rotates amongst strategy / long-horizon sandbox games, so we don't have benchmaxxing..
It's a good direction though, I'll take this into consideration in future versions :D
> Agent messages
> Your AI agents can message you inside the game. A message costs the agent only the words in it, and nothing prints on success. Messages wait while the game is closed.
It's kinda amazing how the ancient "local mail" technology keeps getting reinvented, you know, and reintegrated inside the gameplay? NetHack had its its "mail demon" since... (googles) version 1.0.2, in April 1985 — the creature would spawn in the game, deliver you a scroll of mail, and vanish. And apparently, it's still in use nowadays? "On public servers such as nethack.alt.org, a user may observe other users' games and communicate with them by sending mail."
> Every program attempts to expand until it can read mail. Those programs which cannot so expand are replaced by ones which can.
and yeah, genuinely insane how we came full-circle lol... on terminactory it uses claude code hooks, with an optional "tfac msg" command (tried to minimize token usage, hence the hook)
but yeah, the netHack thing is pretty cool, guess you learn something new every day? xD
To me, it could be interesting for ~10 minutes, and I'd then prefer playing a game like Factorio.
Those aren't competitors.
of course, it's not objectively better than any game, but there are a few separating qualities from the average popular factory game xD
but, now that you mention it, I'll look into adding it soon, thanks!
Will be gpl 3.0 and I plan to release a playable 0.1 on GitHub. I care mostly about historical accuracy of radiation incidents and factory issues at the intersection of nuclear engineering & political issues akin to Paper’s please.
I’m afraid to publish it because I despise the amount of AI slop out there. I just want to tickle my need for a historical plutonium production management simulator.
That was CP437 squeezed into a terminal emulator that ran sixteen or so OpenGL calls per character. I spent a couple months building a shim library to optimise it, before Toady let me fix the game. :p
Real terminal output ought to be faster, but don’t sleep on Kitty.
Especially integrating planners or logic-based approaches with ML models. The old architectures for StarCraft were great at building and micro-managing units but terrible at assessing or adapting to enemy strategy. Hybrid architectures can address that.
Games like yours, esp different styles or requirements, can be used to prove how well the techniques generalize.
It's a good direction though, I'll take this into consideration in future versions :D
It's part of the in-game lore.
> Lovely, a rate limit.
Gitlab rate limits anonymous users. If you sign in, you won't have an issue.