# urantia.dev

> An API and MCP server for the Urantia Papers. Free, with no key.

urantia.dev gives agents and developers structured access to the Urantia Papers: all 197 papers, 1,626 sections, and more than 14,500 paragraphs. You can fetch any paragraph by reference, search by keyword or by meaning, look up named entities, and follow cross-references to the Bible.

## MCP server (for agents)

https://api.urantia.dev/mcp

Streamable HTTP transport. No key and no sign-in. Add it to any MCP client (Claude, Cursor, Windsurf, and others):

{"mcpServers":{"urantia-papers":{"url":"https://api.urantia.dev/mcp"}}}

19 tools: toc.get, papers.list, papers.get, papers.sections, paragraphs.random, paragraphs.get, paragraphs.context, search.fulltext, search.semantic, entities.list, entities.get, entities.paragraphs, audio.get, bible.books, bible.book, bible.chapter, bible.verse, bible.verse.urantia_parallels, bible.search.semantic.

2 resource templates: urantia://paper/{id} and urantia://entity/{id}.
2 prompt templates: study_assistant and comparative_theology.

Recommended flow for an agent:
1. toc.get, to learn the structure.
2. search.semantic for a question, or search.fulltext for exact words.
3. paragraphs.context, to read the paragraphs around a result.
4. papers.get, when the whole paper matters.

Setup guide: https://docs.urantia.dev/mcp-servers
Agent guide: https://docs.urantia.dev/ai-agents

## REST API (for developers)

Base URL: https://api.urantia.dev
OpenAPI 3.1 spec: https://api.urantia.dev/openapi.json
Swagger UI: https://api.urantia.dev/docs
TypeScript client: npm install @urantia/api

No authentication for the endpoints below. The limit is 200 requests per minute per IP address. Errors use RFC 9457 problem+json.

### Text
- GET /toc: table of contents (parts and papers)
- GET /papers: all 197 papers
- GET /papers/:id: one paper with all its paragraphs
- GET /papers/:id/sections: the sections of a paper
- GET /paragraphs/:ref: one paragraph by reference
- GET /paragraphs/:ref/context?window=3: a paragraph with the paragraphs around it (window 1 to 10)
- GET /paragraphs/random: a random paragraph

Add ?include=entities,bibleParallels,urantiaParallels to a paragraph or search request to attach entity mentions and cross-references. Add ?format=rag for a compact shape with a citation and a token count.

### Search
- POST /search or GET /search?q=: full-text search. Body: {"q": "search terms", "type": "and|or|phrase", "limit": 10}
- POST /search/semantic or GET /search/semantic?q=: search by meaning. Body: {"q": "a question in plain words", "limit": 10}
- Both take the filters paperId and partId.

### Entities
- GET /entities: more than 4,400 beings, places, and concepts. Filter with ?q= or ?type=
- GET /entities/:id: one entity
- GET /entities/:id/paragraphs: the paragraphs that name it

### Bible cross-references
- GET /bible/books, GET /bible/:bookCode, GET /bible/:bookCode/:chapter, GET /bible/:bookCode/:chapter/:verse: the World English Bible (81 books)
- GET /bible/:bookCode/:chapter/:verse/urantia-parallels: the 10 nearest Urantia paragraphs for a verse
- POST /bible/search/semantic: search the Bible by meaning, with Urantia paragraphs on each result

The cross-references are semantic neighbors computed from embeddings. They are not a curated list.

### Audio, citations, and more
- GET /audio/:ref: narration URLs for a paragraph
- GET /cite?ref=&style=apa|mla|chicago|bibtex: a formatted citation
- GET /embeddings/:ref and GET /embeddings/export?paperId=: embedding vectors
- GET /og/:ref: a share image for a passage
- GET /languages: available languages
- GET /tools/openai and GET /tools/anthropic: function-calling schemas for those SDKs

### Feedback
- POST /feedback: report a bug, a docs gap, or an idea. No authentication
  - Body: {"category": "bug|docs|api|product|other", "message": "up to 4000 characters"}
  - Optional: ref, endpoint, requestId, client, contact, pageUrl
  - Limit: 5 requests per minute and 10 per 15 minutes per IP address. A 429 has a Retry-After header
  - Guide: https://docs.urantia.dev/feedback

## Paragraph references

Three formats work everywhere and are detected from the string:
- standardReferenceId, paper:section.paragraph, for example 2:0.1
- globalId, part:paper.section.paragraph, for example 1:2.0.1
- paperSectionParagraphId, paper.section.paragraph, for example 2.0.1

## Audio

Every paragraph has narration. Read the `audio` object on the paragraph and use the URLs it gives. Do not build an audio URL yourself: only one voice (tts-1-hd, nova) covers every paragraph.

## About the text

The Urantia Book was published in 1955. Its English text is in the public domain. It has 197 papers in four parts:
- Part I: The Central and Superuniverses (Papers 1 to 31)
- Part II: The Local Universe (Papers 32 to 56)
- Part III: The History of Urantia (Papers 57 to 119)
- Part IV: The Life and Teachings of Jesus (Papers 120 to 196)

The Foreword is Paper 0.

## Links

- Home: https://urantia.dev
- Documentation: https://docs.urantia.dev
- Full documentation as one file: https://docs.urantia.dev/llms-full.txt
- Interactive demo: https://demo.urantia.dev
- Source code: https://github.com/urantia-hub/urantia-dev-api
- Status: https://status.urantia.dev

Sites built with this API include UrantiaHub (https://www.urantiahub.com), a place to read the Papers. They are separate projects that use the API.

urantia.dev is an independent community project. It is not affiliated with Urantia Foundation.