world-builder
Skill39 starsChanged 56 days agoMITView on GitHub
Build the world a story hangs on: characters, locations, factions, rules, and history. Use before or alongside fiction and screenwriting work when the setting needs to be consistent. Runs a layered questioner, generates a structured world bible, and can reverse-map existing drafts into characters, places, factions, rules, timelines, continuity facts, and graph links. Includes Obsidian-ready story folders, character/place notes, fast idea capture, draft cartography, and graph links when the user wants vault integration.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- devkindhq/ideogram-ai-toolkittheir own version, first seen Oct 31, 2026
- Charlie85270/Dorothytheir own version, first seen Oct 25, 2026
- Ultimate-Alchemy/claude-skillstheir own version, first seen Oct 15, 2026
- Samin12/skillstheir own version, first seen Aug 14, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- Charlie85270/Dorothy★ 353same name
- Samin12/skills★ 8same name
- devkindhq/ideogram-ai-toolkit★ 3same name
- Ultimate-Alchemy/claude-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository16 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ewSkill
- auditSkill
- failure-librarySkill
- linkedinSkill
- newsletter-creativeSkill
- newsletter-technicalSkill
- outlineSkill
- sales-copySkill
- scene-structureSkill
- script-writingSkill
- substack-notesSkill
- tweetsSkill
- voiceSkill
- web-copySkill
- workspaceSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 0 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 4 edited versions we read keep these.
0
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 102
When using scene-structure.md or script-writing.md, the world bible is the reference document.
- Very specific: line 118
When Phase 1 is complete, invoke skeleton-template.md to generate the bible.
“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
EW Sub-System — World Builder
Dependency chain. Read these in order before starting:
core/runtime-contract.md(short operating contract; does not replace the full rules)core/obsidian-workspace.md(resolve storage and persistence rules)core/anti-ai-rules.md(full standard)core/ai_slop_commandments.md(mechanism reference and diagnostic checklist)core/voice-profile.md(voice resolver; require the resolved profile'sCompleted:field to be exactlyYes, otherwise stop and run onboarding)- Any
.mdfiles in the active voice's references folder, as resolved bycore/voice-profile.md
Recovery exception: For /ew:world-builder repair <operation-id> or resume, stop this chain after step 4, read obsidian-output.md, and perform only its command-local journal validation/recovery. Do not resolve a voice, load references, ask creative questions, or generate new content.
This sub-system serves the fiction skills. It does not write prose directly. It builds the scaffold the prose will hang on.
WHAT THIS SUB-SYSTEM IS FOR
World-building for fiction: novels, short stories, screenplays, games, and serialized narratives. The world builder extracts a complete, internally consistent world from the writer's ideas and produces a usable world bible — a reference document that makes every writing session faster and every creative decision easier.
The world exists to serve the story. Not the reverse. Every world-building decision made here must answer one question: does this make the story more possible, or does it just make the world more elaborate?
Elaborate is not the goal. Usable is the goal.
THE CORE PROCESSES
Phase 1 — Questioning (questioner.md):
The AI companion conducts a structured, layered dialogue with the writer to extract the world's essential elements. The questions move from the large (what kind of world is this?) to the specific (what does someone eat for breakfast in the capital city?). Not every question applies to every project — the sequence adapts to genre and scope.
Phase 2 — Skeleton Generation (skeleton-template.md):
The AI companion takes the answers from Phase 1 and produces a formatted world bible. Every section of the bible is populated from the questioning. Open questions — things the writer hasn't decided yet — are captured as explicit blanks, not left as gaps that will surprise the writer mid-draft.
Phase 3 — File output:
Read obsidian-output.md only when the user explicitly requests Obsidian graph output and supplies an existing vault root plus a contained story folder. Resolve both component by component, reject symlinks/junctions/reparse redirects, and use vault-root folder-qualified wiki links. A plain or non-vault folder is not graph mode: create at most one portable World Bible.md using ordinary headings, with no frontmatter, wiki links, backlink promises, entity folders, graph indexes, or obsidian-output.md/manuscript-ingest.md graph templates. If the user wants multi-file linked output, ask for an explicit existing Obsidian vault and contained story folder before writing.
Fast Capture Mode — Idea Box to Graph (obsidian-output.md):
When the user gives a small worldbuilding fragment and asks to add, store, capture, remember, or note it, do not run the full questioner. In explicit Obsidian graph mode, read obsidian-output.md, identify the entity and its implied connections, and run its recoverable multi-note transaction protocol across the entity note, 04 - Idea Box.md, hub/index, and all stubs or links. Outside a verified vault, add the fragment under the appropriate ordinary heading in the single portable World Bible.md; do not create entity notes or graph metadata. Ask when the target is unknown or a collision would update the wrong file.
Operation recovery (/ew:world-builder repair <operation-id> or resume):
Dispatch directly to obsidian-output.md's journal recovery procedure. Do not begin the questioner, generate new lore, or run ordinary profile/reference resolution. Perform command-local validation of the explicit vault, story folder, operation ID, journal, and planned paths; resume only the recorded intent. Never create a new operation ID to hide or duplicate an incomplete one.
Draft Cartography Mode - Existing Work to Graph (manuscript-ingest.md):
When the user has already written chapters, scenes, scripts, notes, or a full manuscript and wants the world mapped from the draft, do not start with the questioner. Read manuscript-ingest.md; read obsidian-output.md only for an explicit verified-vault request. Extract entities and relationships from evidence, separate confirmed facts from inferences and open questions, and capture continuity risks. Outside Obsidian graph mode, put the result in the single portable World Bible.md or return it in chat; do not create graph notes.
WHEN TO USE WORLD BUILDER
Use it before writing the first scene whenever possible. Also use it after drafting has already begun. Many writers discover the world by writing into it first; when that happens, the world builder reverse-maps what the draft already implies, then turns those discoveries into a usable bible and graph.
The single most expensive mistake in fiction is leaving the world only inside scattered pages and memory, which means making contradictory decisions in different scenes, then spending revision fixing the inconsistencies instead of improving the prose.
The world bible is a living document. It starts minimal and grows. Ideally it exists before the writing begins. If the writer already has pages, build the first bible from those pages and keep going.
Minimum viable world bible: You do not need to answer every question before you start. You need to answer the questions whose answers will affect the first chapter. Everything else can be developed as you go — but it must be recorded in the bible when it's decided, not trusted to memory.
HOW THE WORLD SERVES THE STORY
The world is not the point. The characters are the point. World-building earns its place when it does one of these things:
- Creates conflict — the world's rules, politics, geography, or history creates a problem the character must navigate
- Reveals character — the world's pressures expose who the character is under stress
- Raises stakes — the world makes winning more meaningful and losing more costly
- Creates specificity — a detailed world makes vague emotional events feel real
World-building that exists only to be admired is a procrastination strategy. If the writer can explain how the magic system works in great detail but cannot answer "what does my protagonist want in chapter one," the world is being used as an escape from the harder problem.
The hard problem is character. The world exists to make the character's problem worse.
THE INTERNAL CONSISTENCY REQUIREMENT
A world only needs to follow its own rules. But it must follow them without exception. The moment the world breaks its own rules for the sake of plot convenience, the reader's trust is broken — and trust, once broken in fiction, is almost impossible to rebuild.
The rule-to-consequence test: Every rule in your world must have a consequence if broken or avoided. A world where magic exists but nothing stops everyone from using it all the time is not a world — it's a decoration. What does magic cost? Who controls it? What happens to people who use too much of it? The constraint is more interesting than the capability.
HOW THE BIBLE INTEGRATES WITH WRITING SESSIONS
When using scene-structure.md or script-writing.md, the world bible is the reference document. Before writing any scene set in a new location, check the bible. Before introducing a new faction, character, or rule — check whether it contradicts anything already established.
When an Obsidian story folder exists, the graph notes are also reference documents. Character, place, faction, object, rule, and timeline notes must be checked before drafting scenes that touch them. If the draft establishes a durable fact, add it to the relevant note and to the continuity log.
Fast capture should feel nearly frictionless. The writer should be able to toss in a niche idea mid-flow and get a linked note back without having to fill out a form.
Draft cartography should feel like switching on the lights inside a manuscript. The writer gives the pages they already have; the world builder returns the characters, places, factions, rules, timeline events, objects, relationships, confirmed continuity facts, contradictions, and open threads that are already present. It must cite where facts came from and avoid inventing missing lore.
The bible is not a creative constraint. It is the creative foundation that makes freedom possible. You can only improvise well when you know the rules deeply.
STARTING THE PROCESS
To begin a new world from fragments, invoke questioner.md. The AI companion will run Phase 1 — the layered question sequence. Bring whatever you have: a premise, an image, a character name, a conflict you know you want to explore. The questioner works with fragments. You do not need a complete vision to begin.
When Phase 1 is complete, invoke skeleton-template.md to generate the bible.
To map an existing draft, invoke manuscript-ingest.md first. Use the draft as the source of truth, then generate a report, update the world bible, or create Obsidian graph files depending on what the user asked for.
For every Obsidian operation whose write set contains multiple notes, preflight and journal the whole set before the first mutation. Mark the operation complete only after re-reading and verifying the graph and indexes.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

