WorldKnowledge (agent)
WorldKnowledge is the agent that handles real-world factual questions. If your fiction takes place on Earth and your character is using a shotgun, WorldKnowledge is the one that checks whether the shotgun works the way you wrote it. If your fantasy world has horses, WorldKnowledge is the one that catches when you wrote something a horse physically cannot do. If your sci-fi novel mentions orbital mechanics, WorldKnowledge is the one that tells you whether your numbers are plausible. And — more subtly — WorldKnowledge is also the agent that manages your project’s world rules: the explicit list of ways your fiction deliberately differs from reality, so that fact-checking respects your intentional breaks instead of flagging them as errors.
Home module
Section titled “Home module”WorldKnowledge has two homes. His primary one is the Chat tab as a selectable agent — you pick him when your question is specifically about real-world facts, not about your own world. His secondary home is the WorldKnowledge tab inside the Legendry, where the Wikipedia search interface and the world rules editor live. The agent uses the same tools the tab exposes, but the agent-via-Chat workflow is more conversational while the tab is more structured.
Ishvana also delegates to WorldKnowledge automatically when she detects a real-world factual question in a chat that isn’t explicitly addressed to any particular agent.
What WorldKnowledge does
Section titled “What WorldKnowledge does”Wikipedia search and retrieval
Section titled “Wikipedia search and retrieval”The foundational capability. WorldKnowledge has access to a local Wikipedia mirror (stored in the project’s data directory) and can run structured searches across it. Give him a query — “how does shotgun shell gauge work” or “life expectancy in medieval Europe” or “how long does it take a horse to travel 40 miles” — and he pulls the relevant Wikipedia article, summarizes the facts, and cites the source.
Unlike generic chatbot responses, his Wikipedia lookups are verifiable. The source article is linked. You can click through and read the actual Wikipedia page to check what he said. If Wikipedia is wrong, you know Wikipedia is wrong. If the agent summarized it correctly, you can trust the summary.
Fact-checking prose
Section titled “Fact-checking prose”Paste a paragraph and ask “is anything in this factually wrong?” and WorldKnowledge runs through it looking for real-world factual claims to check. “The character rode from Grimnest to Garuda in two days on horseback” — WorldKnowledge doesn’t know where Grimnest and Garuda are (that’s your fiction) but if the text implies a real-world distance, he flags the claim. “The character loaded the revolver with seven rounds” — if the revolver is an explicitly real-world model, WorldKnowledge checks the cylinder capacity against reality and flags the mismatch.
The fact-check is advisory. He flags things for your review, not as definitive errors. You get the final call — maybe the character loaded an extra round because of a specific modification, maybe the horse journey is explicitly unusual for a reason. WorldKnowledge surfaces the discrepancy; you decide what to do with it.
World rules
Section titled “World rules”This is the feature that makes WorldKnowledge genuinely useful for fantasy writers. Most fiction that isn’t strictly realistic has some rules that explicitly differ from reality — magic exists, the sun works differently, horses are smarter than real horses, whatever. Without some way to tell the agent about these rules, every fact-check would flood you with false positives against things you deliberately changed.
World rules are explicit axioms you define for your project. Each rule is a short statement of the form “in this fiction, X is true (even though X is not true in reality).” Examples:
- “In this fiction, magic exists and can be used to accelerate wound healing. Characters can recover from injuries faster than real-world medicine would allow.”
- “In this fiction, horses can travel significantly further in a day than real-world horses, because the Tikor strain of horse has been magically enhanced.”
- “In this fiction, the sun rises in the west and sets in the east.”
- “In this fiction, gunpowder has been suppressed by magical fiat. No firearms exist regardless of otherwise-realistic technology levels.”
When WorldKnowledge runs a fact-check on your prose, he first consults the world rules. If a prose claim contradicts reality but matches a world rule, he treats it as correct. If a prose claim contradicts both reality and the world rules, he flags it as a genuine error.
This means you can run fact-checking on a fantasy manuscript without getting spammed about magic. You told the agent what the rules are. He respects them.
World rules management
Section titled “World rules management”You define and edit world rules from the WorldKnowledge tab in the Legendry module. The tab has a dedicated rules editor — a list of active rules, an input to add new ones, severity levels, tags, and a preview of what the agent sees when he consults them.
Rules can be project-scoped (this specific book’s rules) or project-wide (rules that apply to every book in the project). The distinction matters because sometimes you have a series where each book has a slightly different magic system, and you want the book-specific rules to override the series-wide ones for that book.
What the agent is actually for
Section titled “What the agent is actually for”A few specific use cases.
Historical fiction accuracy. You’re writing a Regency-era romance and you want to know whether a specific phrase was in use yet, or whether a particular garment was worn by women of that class, or whether a tool existed yet. WorldKnowledge pulls the relevant Wikipedia article and tells you.
Modern thriller plausibility. You’re writing a thriller where the protagonist needs to do something specific — hotwire a car, pick a lock, disable an alarm system — and you want to know how it actually works so the scene doesn’t read as made up. WorldKnowledge finds the sources.
Fantasy grounding. Your fantasy world has real-world horses. You want to know how fast they actually travel, how much they eat, how long they can work before they need rest. WorldKnowledge gives you the real-world baseline so your fictional horses feel grounded.
Sci-fi plausibility. You’re writing hard science fiction and you want your orbital mechanics, physics, chemistry, or biology to be defensible. WorldKnowledge checks against real science and flags things that are wrong. You can then decide whether to fix the prose or add a world rule that explains why the physics works differently.
What WorldKnowledge isn’t
Section titled “What WorldKnowledge isn’t”A few things this agent deliberately doesn’t do.
He’s not a scholar. WorldKnowledge’s source is Wikipedia (and the world rules). He’s not cross-referencing peer-reviewed journals, he’s not consulting primary sources, he’s not looking at academic papers. Wikipedia is pretty good, but it’s not perfect, and for anything that matters you should double-check with real sources.
He doesn’t know your world. WorldKnowledge’s job is real-world facts. If you ask him about your own characters or your own cultures, he’ll redirect you — “that sounds like a question for Ishvana or the Lorekeeper.” The division of labor is real: WorldKnowledge handles what’s outside your fiction, the other agents handle what’s inside it.
He’s not an opinion machine. If you ask WorldKnowledge “is this the right historical period for my story,” he won’t give you a craft opinion. He’ll tell you facts about the historical period and let you decide.
Interaction with the Lorekeeper
Section titled “Interaction with the Lorekeeper”There’s a subtle but important division of labor between WorldKnowledge and Lorekeeper.
Lorekeeper checks your prose against your own canon. Did the character’s eye color change between chapters? Did the city name drift? Did a promised plot thread fail to resolve? These are internal-canon questions and Lorekeeper handles them.
WorldKnowledge checks your prose against reality (as modified by your world rules). Did the character do something a real person couldn’t do? Is a real-world fact wrong? These are external-reality questions and WorldKnowledge handles them.
In a perfect audit, you’d run both — Lorekeeper for internal consistency, WorldKnowledge for external plausibility, and between the two your prose is tight on both dimensions. Most authors run Lorekeeper more frequently because internal contradictions are more common and higher-stakes. WorldKnowledge becomes more valuable as your manuscript gets closer to submission and you’re looking for the kind of “wait, that wouldn’t actually work” critique a careful beta reader might catch.