_Michael_
Explorer
World Foundry — worldbuilding that starts you somewhere instead of nowhere
I've been building a tool called World Foundry, and I'd rather have it argued with by people who actually build worlds than keep refining it in private.
Here's the problem I kept hitting. The tools available now split into two camps. Map generators give you a beautiful map with nothing behind it — Azgaar, Wonderdraft, watabou's generators are all excellent at what they do, and what they do is produce an image. Wikis go the other way: World Anvil hands you a well-organized set of empty fields and waits. Both are good tools. Both leave the actual work — the thousand small coherent decisions that make a place feel real — entirely on you.
So a new worldbuilder gets a gorgeous coastline and a blank page, and the blank page wins.
World Foundry generates downward, and each level inherits from the one above it. You zoom continuously — world, nations, cities, villages, eventually individual buildings — and every level you enter gets generated and permanently recorded in a wiki you own. A village doesn't come out of a random table. It comes out of what's actually in that region: building materials from local resources, distributed the way real resources are. Architectural sophistication scaled by population and artisan presence, with a floor set by whatever the ruling authority has standardized — so a modest settlement in a well-run realm looks modest, not primitive. Even population distributions of NPC types (artisans, blacksmiths, vendors, etc) and different currencies and exchange rates will be supported.
You're not authoring from nothing. You're editing something coherent. That's the entire difference.
And the editing is the point, not an afterthought. Change the flavor text however you like — freeform, or quick actions to expand or condense — and the system tracks what you changed and why, tuning toward the cultural texture of your world rather than word count. Correct it once, and it stops making that mistake.
Some commitments, since they matter more than the features:
OGL 1.0a only. No dependency on 1.1 or the VTT policy. Likely Product Identity terms replaced with my own.
Open source, MIT-licensed throughout.
Nothing scraped. Point it at your own reference library and it stays local to you.
You approve every update. Nothing installs itself, and anything unfamiliar comes with a plain-language explanation of what it changes.
The AI layer — I call it the Chamberlain — is deliberately swappable (use your own LLM API!), and it does the drafting, not the deciding. Lighter local models handle the mechanical work so the expensive calls are reserved for things that actually need judgment.
It's being built against a real setting rather than a test case. Atlán is a low-magic world I've been developing for years, and it ships free and complete with the engine: a working example, and something people can add to. Contributions that fit the world become canon, with credit.
Early development, nothing to download yet. But the decisions above are locked in, and I'd rather hear the objections now. World Foundry is a working name and may change. No affiliation with Foundry Virtual Tabletop or Foundry Gaming LLC. Figured I'd get that out of the way--open to some cool names for it if you think you can do better.
The goal is a product for $49.95 that comes with an entire low-magic game world already present and waiting to be activated (or not used if you are building your own) that doesn't require mass deletions to prune out. The material for the campaign setting would be over 1000+ pages in print, so hundreds of dollars worth of material for free bundled with it. We may also have a second world that comes with it as an option as well. It's meant to be easy to use for first time RPG players and world builders, and easily transportable to your games via a thumb drive. Plug in, fire it up, and you have an instant wiki of your world at your finger tips.
Would love to hear any feedback you have to make it better ultimately for everyone.
I've been building a tool called World Foundry, and I'd rather have it argued with by people who actually build worlds than keep refining it in private.
Here's the problem I kept hitting. The tools available now split into two camps. Map generators give you a beautiful map with nothing behind it — Azgaar, Wonderdraft, watabou's generators are all excellent at what they do, and what they do is produce an image. Wikis go the other way: World Anvil hands you a well-organized set of empty fields and waits. Both are good tools. Both leave the actual work — the thousand small coherent decisions that make a place feel real — entirely on you.
So a new worldbuilder gets a gorgeous coastline and a blank page, and the blank page wins.
World Foundry generates downward, and each level inherits from the one above it. You zoom continuously — world, nations, cities, villages, eventually individual buildings — and every level you enter gets generated and permanently recorded in a wiki you own. A village doesn't come out of a random table. It comes out of what's actually in that region: building materials from local resources, distributed the way real resources are. Architectural sophistication scaled by population and artisan presence, with a floor set by whatever the ruling authority has standardized — so a modest settlement in a well-run realm looks modest, not primitive. Even population distributions of NPC types (artisans, blacksmiths, vendors, etc) and different currencies and exchange rates will be supported.
You're not authoring from nothing. You're editing something coherent. That's the entire difference.
And the editing is the point, not an afterthought. Change the flavor text however you like — freeform, or quick actions to expand or condense — and the system tracks what you changed and why, tuning toward the cultural texture of your world rather than word count. Correct it once, and it stops making that mistake.
Some commitments, since they matter more than the features:
OGL 1.0a only. No dependency on 1.1 or the VTT policy. Likely Product Identity terms replaced with my own.
Open source, MIT-licensed throughout.
Nothing scraped. Point it at your own reference library and it stays local to you.
You approve every update. Nothing installs itself, and anything unfamiliar comes with a plain-language explanation of what it changes.
The AI layer — I call it the Chamberlain — is deliberately swappable (use your own LLM API!), and it does the drafting, not the deciding. Lighter local models handle the mechanical work so the expensive calls are reserved for things that actually need judgment.
It's being built against a real setting rather than a test case. Atlán is a low-magic world I've been developing for years, and it ships free and complete with the engine: a working example, and something people can add to. Contributions that fit the world become canon, with credit.
Early development, nothing to download yet. But the decisions above are locked in, and I'd rather hear the objections now. World Foundry is a working name and may change. No affiliation with Foundry Virtual Tabletop or Foundry Gaming LLC. Figured I'd get that out of the way--open to some cool names for it if you think you can do better.
The goal is a product for $49.95 that comes with an entire low-magic game world already present and waiting to be activated (or not used if you are building your own) that doesn't require mass deletions to prune out. The material for the campaign setting would be over 1000+ pages in print, so hundreds of dollars worth of material for free bundled with it. We may also have a second world that comes with it as an option as well. It's meant to be easy to use for first time RPG players and world builders, and easily transportable to your games via a thumb drive. Plug in, fire it up, and you have an instant wiki of your world at your finger tips.
Would love to hear any feedback you have to make it better ultimately for everyone.
