# Middle-earth Wiki

> Agent-facing counterpart to the [human project page](/projects/fictional-world-llm-wiki/).

## Record metadata

- Record: 024
- Slug: fictional-world-llm-wiki
- Domain: Creative
- Domain code: CRE
- Type: Knowledge system
- Status: Active
- Period: 2026
- Portfolio role: Concept demonstration
- Publication state: Public methodology; copyrighted text excluded
- Case-study readiness: Draftable
- Compendium edition: 0.4

## Summary

A grounded LLM wiki for fictional worlds, using layered truth to separate canon, interpretation, adaptation, campaign, and private invention.

## Overview

Fictional worlds create a grounding problem that ordinary search handles poorly. Canon, adaptations, interpretation, role-playing campaigns, and private invention can all sound equally authoritative when an AI model blends them into one confident answer.

A Middle-earth-based demonstration of a broader LLM-wiki architecture for books, films, television, games, and original worlds.

Purpose: Make rich fictional settings easier for people and AI agents to explore without quietly blending incompatible sources.

## The problem behind the project

Ordinary wikis flatten uncertainty and adaptation differences, while general language models can invent plausible lore. Layered provenance gives agents a grounded place to reason from.

The Middle-earth Wiki experiment treats 'what kind of true is this?' as data. A claim can belong to canon, a particular adaptation, an interpretation, a campaign, or a private layer, allowing a reader or agent to reason without silently crossing those boundaries.

Readers, players, writers, game masters, lore researchers, and AI agents may benefit. Authors and rights holders are affected, so the public system must respect copyrighted source material.

## How it took shape

Local structured notes and an agent-readable navigation model that distinguishes canon, interpretation, adaptation, campaign-specific material, and private additions.

The private prototype uses structured notes and an LLM-readable navigation system rather than a copied corpus. Middle-earth is the motivating example, but the architecture is intended for books, films, television, games, and original worldbuilding.

Josiah developed the LLM-wiki pattern, selected fictional worlds as a high-value use case, and explored it through a Middle-earth knowledge base.

A working private knowledge-base experiment demonstrates the navigation and layering pattern; no complete public lore corpus is offered.

## What the project means now

The most promising public form is not an unofficial franchise encyclopedia filled with copyrighted text. It is a replicable grounded-wiki method demonstrated with original or public-domain material, showing how provenance can make lore useful to both people and agents.

Copyrighted text, private campaign material, and unsupported lore claims cannot be published. Grounding quality depends on careful source work.

Fictional knowledge becomes more useful to agents when 'what kind of true is this?' is encoded alongside the fact.

Publish the architecture with a small original or public-domain demonstration world that others can replicate safely.

## Publication and interpretation notes

- Current classification: Active
- Portfolio readiness: Draftable
- Publication boundary: Public methodology; copyrighted text excluded

## Additional agent context

Middle-earth is the motivating example, not a promise to republish Tolkien text. Focus on the generalized layered-truth architecture.

## Related project records

- [Personal Context Repository](/projects/personal-context-infrastructure/llm/) — A human-owned, agent-readable knowledge system that makes long-running AI collaboration more coherent without surrendering control of the record.
- [MasteryGraph](/projects/masterygraph/llm/) — A learning methodology that treats knowledge as an inspectable graph of claims, prerequisites, evidence, and mastery.

## Navigation

- [Complete project index](/projects/llm/)
- [Human version of this record](/projects/fictional-world-llm-wiki/)
- [About Josiah's working method](/about/llm/)
- [Agent discovery map](/llms.txt)