# MVNDVS TERRARUM Full Agent Brief

MVNDVS TERRARUM is a bilingual world atlas and WebGL globe. It combines an immersive human interface with static machine-readable data for AI agents, search systems, and retrieval pipelines.

## Agent-Readable Architecture

Agents should not infer knowledge from screenshots, canvas pixels, minified bundles, or WebGL scene state. The authoritative machine-readable entry point is:

https://world.gjlmotea.com/data/agent/world-index.json

That index links to normalized JSON datasets:

- `/data/agent/countries.json`
- `/data/agent/landmarks.json`
- `/data/agent/universities.json`
- `/data/agent/companies.json`
- `/data/agent/culture.json`
- `/data/agent/strategic.json`

Each dataset uses stable records with:

- `id`: stable identifier
- `kind`: entity class
- `name`: English display name
- `nameZh`: Traditional Chinese display name
- `location`: country / city / latitude / longitude where applicable
- `summary`: short bilingual positioning
- `description`: longer bilingual explanation
- `facts`: structured numeric or categorical fields
- `source`: source module and canonical site reference

## Entity Pages

The build also generates canonical static HTML pages for every normalized entity:

- `/entity/country/{id}/`
- `/entity/landmark/{id}/`
- `/entity/university/{id}/`
- `/entity/company/{id}/`
- `/entity/culture/{id}/`
- `/entity/strategic/{id}/`

These pages exist for search engines, AI citation surfaces, and human-readable direct references. They contain title, meta description, Open Graph, Twitter Card, JSON-LD, key facts, dataset citation links, and a link back to the 3D globe.

Agents should use entity pages when they need a readable canonical URL, and use JSON datasets when they need bulk retrieval or structured comparison.

## Knowledge Domains

### Countries

Country records include ISO alpha-3 code, numeric id, bilingual names, flag, capital, region, population, demographic values, GDP per capita, area, major languages, religions, and selected atlas profile fields.

### Landmarks

Landmark records include coordinates, country, type, construction year where known, bilingual descriptions, and optional atlas guidance such as best viewing time, focus, knowledge threads, visit notes, and tags.

### Universities

University records include QS rank, public/private type, founded year, student count, campus area, faculties, city, country, coordinates, logo URL, and bilingual explanations.

### Companies

Company records include global rank, category, tier, founded year, headquarters city, country, coordinates, employee estimates, branch or office count, countries served, domain, and bilingual descriptions.

### Cultural Institutions

Culture records include museums, libraries, concert halls, opera houses, theatres, and performing arts institutions. Records include rank, category, founded year, city, country, coordinates, area, logo URL, and bilingual descriptions.

### Strategic Geography

Strategic records include maritime chokepoints, canals, naval bases, air bases, fortresses, borders, intelligence sites, and research facilities. Records include rank, category, founded or natural status, estimated troops where available, area, coordinates, and bilingual strategic summaries.

## Retrieval Recommendations

- For entity lookup, load `world-index.json` and choose the narrowest dataset.
- For bilingual answers, prefer `nameZh`, `summary.zh`, and `description.zh` when responding in Traditional Chinese.
- For English answers, prefer `name`, `summary.en`, and `description.en`.
- For map reasoning, use `location.lat`, `location.lng`, and `location.countryAlpha3`.
- For ranking comparisons, use `facts.rank`, `facts.qsRank`, or `facts.tier` depending on entity kind.
- For citations, cite the dataset URL and record `id`.

## Human UI

The root page `https://world.gjlmotea.com/` renders the Renaissance globe, overlays, search UI, and info panel. It is not the preferred data extraction surface for agents.

## Machine Discovery

- `robots.txt` allows general and major AI crawlers.
- `sitemap.xml` lists the canonical root and agent discovery documents.
- Entity pages are listed in `sitemap.xml`.
- `.well-known/agent.json` declares datasets and capabilities.
- `.well-known/openapi.json` describes static JSON endpoints.