# Bible Research MCP Server

A local [Model Context Protocol](https://modelcontextprotocol.io/) server for reading, searching, researching, and studying the Bible. It uses only Node.js built-in modules: there are no npm runtime or development dependencies, no network requests, and no external database.

The included corpus contains 66 books, 1,189 chapters, and 31,104 verses. The server reads the Markdown files in `books/` when it starts.

## Requirements

- Node.js 18 or newer

## Run it

```sh
npm start
```

MCP clients normally launch the process themselves. For a client that accepts JSON configuration, add something equivalent to:

```json
{
  "mcpServers": {
    "bible": {
      "command": "node",
      "args": ["/absolute/path/to/bible/server.js"]
    }
  }
}
```

The stdio transport uses one UTF-8 JSON-RPC message per line. Standard output is reserved for protocol messages; diagnostics go to standard error.

## Tools

- `get_passage` — read a chapter, verse, range, or cross-chapter range.
- `search_bible` — literal `all`, `any`, or `phrase` search with book/testament filters.
- `research_topic` — related terms, distribution by book, and representative passages.
- `word_study` — English concordance counts and verse contexts.
- `find_related_passages` — transparent TF/IDF-like lexical similarity discovery.
- `compare_passages` — exact texts plus shared and distinctive terms.
- `get_bible_overview` — corpus statistics and the full book catalog.

Examples of accepted references include `John 3:16`, `Jn 3:16-18`, `Genesis 1-2`, `Genesis 1:31-2:3`, and `1 John 1:1`. Errors are returned as tool results so an AI client can correct a reference and retry.

## Resources and prompts

The server exposes a catalog resource, one Markdown resource per book, and a `bible://passage/{reference}` resource template. It also includes prompts for passage study, topical research, and passage comparison. The prompts explicitly distinguish textual observation from interpretation and ask the client to cite its evidence.

`find_related_passages` is intentionally described as lexical discovery, not as an authoritative cross-reference system. `word_study` analyzes the included English text and does not claim to perform Hebrew or Greek lemma analysis.

## Test

```sh
npm test
```

The tests use Node's built-in test runner and exercise both the corpus API and a real stdio MCP session.