Worldbuilder’s Toolkit
Digital workspaces for managing massive fictional universes
A world of a thousand entries and several authors fails differently from a small one. Solo and small, the problem is recall. Large and shared, the problem is authority: two people making incompatible decisions in good faith, each believing the question was settled. Tooling helps only if it records who decided what, and when.
What breaks at scale
Search stops being enough. Under a couple of hundred entries, search covers for weak structure. Above it, the failure is that you cannot tell whether the thing exists at all. You search “harbour city”, find nothing, invent one, and discover the duplicate eleven months later.
The fix is typed entries and enforced links, so a place cannot end up existing twice under two spellings.
Undeclared canon multiplies. With one author, “settled versus provisional” lives in a head. With four, it does not exist unless it is written on the entry.
Permissions collide with secrecy. In a large shared world some collaborators should not see some entries. Most tools treat permission as a per-world switch; what you need is per-entry.
Leaving becomes expensive. At a thousand entries, export stops being theoretical. Check it before you commit, not after.
What to look for
| Requirement | Why it matters at scale |
|---|---|
| Typed entries | Prevents the same thing existing twice |
| Per-entry permissions | Secrets are per-fact, not per-world |
| Canon status | Stops good-faith contradiction |
| Real export | A thousand entries is not re-typable |
| Seat pricing | The usual hidden cost, and where tools diverge most |
The platforms
vvd
The Studio plan is $29.17/month or $350/year, covering five collaborators with further seats at $5.83/month each, plus roles and permissions, an admin panel and offboarding. Pro at $10.75/month covers three collaborators and adds API access, which is what makes bulk export and external tooling possible at scale. (Verified 22 September 2026.)
The distinguishing feature for a serious shared project is that worlds on that plan are owned by the studio rather than by whoever created the account. Most tools tie a world to its creator, which is a problem nobody notices until a collaborator leaves.
World Anvil
The most mature permission system in the category, with co-authors, reader groups and per-article control refined over years. If your universe is large, public, and has an audience, this is the one with the most road behind it.
Co-author seats are what matter at scale, and they scale with the tier: Master at $7/month gives 4, Grandmaster at $12/month gives 9, and Sage at $34/month gives 20, all billed monthly and excluding VAT. Sage also lifts subscribers to 1,000 and campaigns and worlds to unlimited. (Verified 22 September 2026.)
The cost is the same as ever: it is the steepest learning curve here, and at scale that cost is paid by every new collaborator you onboard.
Obsidian
Handles size better than people expect, since thousands of Markdown files search instantly and everything is yours. Collaboration means Git or a synced folder, which works for two technical people and does not work for six who are not.
Free for personal use; commercial licence $50/user/year. (Verified 22 September 2026.)
Kanka
Open-source, permissions available, generous free tier. The right answer for a large group that wants shared structure at no cost and can live with a functional interface.
LegendKeeper
Sharing administration is straightforward. Free Basic; Pro $9/month, or $7.50/month billed annually. (Verified 22 September 2026.)
It has been around a long time without ever becoming the obvious answer. At scale it felt like a thin layer over Obsidian or Notion, and a single paid tier is awkward for a team besides, where per-seat pricing elsewhere fits a growing group better.
The conventions that matter more than the tool
At scale these determine success more than any feature.
One owner per area. Per region, per faction, per era, rather than per entry. Ambiguous ownership is what produces contradictory decisions, and no permission system fixes it.
Canon status, visible on every entry. Settled, provisional, abandoned.
A dated decision log. One append-only page: what was decided, by whom, what it overrides. Almost nobody keeps one, and a line per decision ends the argument about whether something was settled.
Name the arbiter before you need one. The moment two collaborators disagree about canon is the worst possible time to negotiate who decides.
A naming convention. At a thousand entries, “the Compact” and “Compact, the” and “The Compact (faction)” will all exist unless someone wrote down which is right.
Frequently asked
What is the best platform for a very large fictional universe?
At scale the deciding features are search quality, per-entry permissions and export. World Anvil and vvd both hold up; Obsidian handles size well but needs a plugin stack. The tool matters less than having canon status on every entry and one owner per area.
How many entries before a world needs real tooling?
Around two hundred. Below that, search covers for structure. Above it, you stop being able to remember whether something exists, and a flat search cannot tell you whether two entries describe the same thing.
How do multiple authors avoid contradicting each other?
Permissions prevent accidents; an owner per area prevents disagreements. You need both, plus a dated decision log. Most shared-world failures are two people deciding in good faith that a question was already settled.
What happens to a shared world if a collaborator leaves?
It depends on who owns the account, which is worth checking before you start. Most tools tie the world to its creator. vvd is a notable exception, since worlds on its Studio plan are owned by the studio, with an offboarding flow.