AETHERION ARCHIVE — Product Requirements Document Aetherion Archive is an AI-native fantasy universe engine that generates, maintains, and serves a living encyclopedia of a dark fantasy world. It is not a blog, not a wiki, and not an article generator — it is a semantic entity graph that renders as a static website.
The core premise: after the Celestial Fracture shattered the moon, fragments fell to the world, becoming the source of all magic — but every spell cast erodes the caster's soul.
Existing fantasy wikis (Fandom, Wikia) are:
Human-curated and slow to expand
SEO-hostile due to ad-heavy platforms
Inconsistent by nature (multiple editors, no canon enforcement)
Not designed for discovery graph optimization
Aetherion Archive solves this by:
AI-generated entity expansion with canon enforcement
Graph-native internal linking (not keyword-stuffed)
Static-first, infinitely scalable deployment
Topical authority through semantic clustering
Dimension
Traditional Fantasy Wiki
Aetherion Archive
Content source
Human editors
AI + canon validation
Internal linking
Manual, inconsistent
Graph-derived, automatic
Lore consistency
Varies by editor
Enforced by rules engine
Expandability
Requires human effort
Automated pipeline
SEO strategy
Afterthought
Graph-first architecture
Deployment
CMS/server
Static (Cloudflare Pages)
Scale ceiling
Database/bandwidth
CDN edge (unlimited)
Primary: SEO Content Consumer
Searchers for dark fantasy / worldbuilding content
TTRPG players seeking campaign inspiration
Fantasy readers exploring new lore
Discovered via long-tail SEO queries
Secondary: Worldbuilding Community
Writers and game masters
People who explore fictional universes
Referenced by other worldbuilding sites
Tertiary: AI/Engineering Audience
Developers interested in AI content pipelines
People studying semantic SEO at scale
Open source contributors
"I want to learn about the Kingdom of Eldoria and its gods"
"Show me how the Celestial Fracture connects to this artifact"
"Give me related entities I haven't explored yet"
"Let me browse all spells in the universe"
"Index this entity page and understand its topic"
"Follow internal links to discover related content"
"Understand the topical authority of the domain"
"Surface this page for relevant fantasy queries"
"Generate 3 new entities focused on the Northern Kingdoms"
"Validate canon consistency before deploying"
"Show me which entity types are underrepresented"
"Rerun the link freshness pass"
4. FUNCTIONAL REQUIREMENTS
ID
Requirement
Priority
F-01
System shall support 10 entity types: kingdoms, factions, races, gods, artifacts, spells, events, monsters, cities, religions
P0
F-02
Each entity shall have structured metadata (type, name, aliases, relationships, attributes)
P0
F-03
Each entity shall have a relationship array linking to other entities
P0
F-04
Entities shall support lifecycle states: active, deprecated, rewritten, archived
P0
F-05
System shall prevent naming collisions via naming registry
P0
F-06
System shall support per-type JSONL storage for independent generation
P0
4.2 Core — World Memory Engine
ID
Requirement
Priority
F-07
World memory shall track current era, cosmic state, faction influence, major conflicts
P0
F-08
World memory shall be append-only journal with snapshot history
P0
F-09
World memory shall be updated after every entity generation
P0
F-10
World memory shall be included in every AI generation prompt
P0
ID
Requirement
Priority
F-11
System shall support multiple AI providers: OpenAI, Gemini, Groq, OpenRouter, OpenAI-compatible
P0
F-12
System shall implement provider abstraction via AIProvider interface
P0
F-13
System shall implement priority-based fallback chain
P0
F-14
System shall assemble context from canon before generation
P0
F-15
System shall validate generated content against canon rules
P0
F-16
System shall weight entity type selection by distribution gap
P0
F-17
System shall support thematic generation seeds / regional focus
P1
ID
Requirement
Priority
F-18
System shall generate a fully static site via Astro
P0
F-19
Every entity shall have a dedicated page at /[type]/[slug]
P0
F-20
Every page shall include internal links from its relationship array
P0
F-21
Every page shall include Schema.org JSON-LD structured data
P0
F-22
System shall generate a sitemap organized by entity type
P0
F-23
System shall have a timeline page
P1
F-24
System shall support TailwindCSS theming
P0
ID
Requirement
Priority
F-25
Internal links shall originate from defined relationships only (lore-first, not keyword-first)
P0
F-26
All links shall be bidirectional at the graph level
P0
F-27
System shall organize entities into topical clusters
P0
F-28
Each entity type shall have a listing page
P0
F-29
Sitemap shall be split by entity type for crawl efficiency
P0
ID
Requirement
Priority
F-30
GitHub Actions shall run daily entity generation
P0
F-31
GitHub Actions shall validate canon on every PR
P0
F-32
GitHub Actions shall build and deploy to Cloudflare Pages on main
P0
F-33
Weekly maintenance shall refresh internal links
P1
F-34
Generation plan shall balance entity type distribution
P0
5. NON-FUNCTIONAL REQUIREMENTS
ID
Requirement
Target
NF-01
Static build time (300 entities)
< 2 minutes
NF-02
Page load time
< 500ms (CDN cached)
NF-03
Lore consistency violations
< 1% of generated entities
NF-04
Internal link density
≥ 5 per entity page
NF-05
Entity types in balance
±20% of target distribution
NF-06
AI generation success rate
> 90% (with fallback)
NF-07
Zero runtime server dependencies
Static files only
NF-08
Incremental build support
Changed entities only
Request → Generate → Validate → Canon Store → Content Render → Build → Deploy
┌─────────────────────────────────────┐
│ ENTITY STATES │
│ │
│ draft ──→ active ──→ deprecated │
│ │ │ │
│ ↓ ↓ │
│ rewritten ──→ archived │
└─────────────────────────────────────┘
draft : Being generated, not yet in canonactive : Live in canon, rendered on sitedeprecated : Superseded; page redirects to replacementrewritten : Content replaced; old version stored in memory journalarchived : Removed from graph; preserved for history
Fully static deployment (Cloudflare Pages)
No runtime database (JSONL canon)
Zero JavaScript on entity pages (progressive enhancement only)
Markdown + JSONL for all content
Git-tracked canon with full history
Must originate from the Celestial Fracture as central event
Magic must always have a soul cost
No generic fantasy tropes (anti-slop rules enforced)
Every entity must connect to at least 3 others
Names must be unique across the entire canon
AI provider keys loaded from environment only, never committed
PRs from auto-generation must be reviewed before merge
Canon validation runs on every change
Weekly link freshness pass required
Metric
Goal
Measurement
Entity count
300+ entities launched
Generated canon count
Internal links per entity
≥ 5
Linker pass output
Entity type balance
All types within 20% of target
Distribution JSON
Build time
< 2 min
CI pipeline duration
Lore conflicts
< 3 per 100 entities
Validator warnings
Generation success
> 90%
Pipeline pass rate
Phase 1 — Foundation (Current)
Canon reader/writer
Entity types + JSONL storage
World Memory Engine
Naming registry
Provider abstraction layer
Context assembler + generator pipeline
Validator (warning-based)
Renderer (entity → markdown)
Astro static site (entity pages, listing pages, timeline)
SEO: Schema.org, sitemap, breadcrumbs
GitHub Actions: daily seed, validate, build-deploy
Seed: 30-50 entities across all types
Thematic generation seeds / focus campaigns
Graph visualization (D3.js)
Client-side search
Content relationship graph page
Advanced SEO: pillar/cluster optimization
Entity diff viewer for PR review
Million-entity graph optimization
Dynamic sitemap federation
Multi-language support
Community contribution system
API for external consumption
Term
Definition
Entity
A single node in the universe graph (kingdom, god, spell, etc.)
Canon
The complete, authoritative body of lore
World Memory
The living state of the universe at a given point
Relationship
A directed edge between two entities with a semantic label
Topical Cluster
A group of entities forming an SEO-relevant content group
Anti-slop
Rules preventing generic, trope-filled AI generation
JSONL
JSON Lines — one JSON object per line, append-friendly