Gospel Library MCP Server (Unofficial)

This is a Model Context Protocol server for LLM interfaces such as Claude, ChatGPT, or LM Studio. You can set up a remote server via Cloudflare Workers (free), or you can run the MCP server locally using node.

Quick Deploy

One-click deployment to Cloudflare Workers - Sets up your own free MCP server instance in under 2 minutes.

Tools Overview

Current active tools:

1. get_exact_scripture - Fetch an exact LDS scripture verse or short contiguous range

   • reference (string, required): A verse or short range like "John 3:16", "Alma 32:27-28", "1 Nephi 3:7". Range limit: ≤50 verses.
   • Fuzzy matching supported: "Matt 5:16" → "Matthew 5:16", "Omni 7" → "Omni 1:7", "Doc 121" → "D&C 121"
   • Always call before quoting scripture wording to ensure accuracy.

2. search_scriptures_by_keyword - Search LDS scriptures by keyword/phrase for topic discovery

   • query (string, required): Keyword or short phrase (<100 chars), e.g. "charity", "plan of salvation", "endure to the end"
   • limit (number, optional): Max number of results (default 10, max 20)
   • Use before teaching on a topic or when user asks "verses about X".

3. get_random_scripture - Return a single random scripture verse

   • No parameters required
   • Useful for daily verse prompts or inspiration.

4. search_conference_talks - General Conference talks search and retrieval

   • id (number, optional): Specific talk ID to retrieve full talk
   • query (string, optional): Keyword(s)/phrase to search in talk content (<100 chars)
   • speaker (string, optional): Speaker name with fuzzy matching. E.g. "Nelson", "Russell M. Nelson", "Russel Nelson"
   • conference (string, optional): Conference identifier with fuzzy matching. E.g. "April 2023", "Oct 2022", "October 1990"
   • limit (number, optional): Maximum results (default 10, max 20)

   Enhanced Features:

   • Fuzzy matching: "Russel M. Nelson" matches "Russell M. Nelson", "Oct 1990" matches "October 1990"
   • Smart single results: When only 1 talk matches, returns full talk content automatically
   • Intelligent filtering: Combines multiple parameters for precise searches
   • Always fetch before quoting to ensure accuracy.

Typical Query Flows

Find a specific scripture reference:

get_exact_scripture{ reference: "John 3:16" }
get_exact_scripture{ reference: "Omni 7" }  // Fuzzy matches to Omni 1:7
get_exact_scripture{ reference: "Matt 5:16" }  // Fuzzy matches to Matthew 5:16

Search scriptures by topic:

search_scriptures_by_keyword{ query: "charity", limit: 5 }
search_scriptures_by_keyword{ query: "plan of salvation" }

Find a Russell M. Nelson talk from October 1990:

search_conference_talks{ speaker: "Russell M. Nelson", conference: "October 1990" }

Search for talks about a specific topic:

search_conference_talks{ query: "atonement", speaker: "Russell M. Nelson", limit: 5 }
search_conference_talks{ query: "faith", conference: "April 2023" }

Get a full talk by ID:

search_conference_talks{ id: 12345 }

Get a random scripture:

get_random_scripture{}

Key Features

Scripture Tools:

• Comprehensive coverage: Bible, Book of Mormon, Doctrine & Covenants, Pearl of Great Price
• Intelligent fuzzy matching: Handles abbreviations and common misspellings
• Flexible reference parsing: Supports "Book Chapter:Verse", "Book Verse" (assumes chapter 1), and D&C sections
• Fast keyword search: Full-text search across all scriptures
• Random verse generator: Perfect for daily inspiration

Conference Talk Tools:

• Modern prophet teachings: Conference talks from April 1971 onward
• Smart speaker matching: Finds speakers even with typos ("Russel" → "Russell")
• Conference abbreviations: "Oct 1990" → "October 1990", "Apr 2023" → "April 2023"
• Intelligent results: Single matches return full talk content automatically
• Multi-parameter filtering: Combine speaker, conference, and keyword searches

Error / Guidance Messages

The tools return actionable hints when no results are found, suggesting alternative search approaches or parameter adjustments.

Local Development

Type check:

bun run type-check

Deploy:

wrangler deploy

Tail logs:

wrangler tail

Local Database Priority (Local -> D1)

At runtime the server resolves the database in this order:

1. Local file gospel-library.db (Node using better-sqlite3 if available, otherwise Bun's built-in sqlite)
2. Cloudflare D1 binding DB (only if no local file connection was established)

Place a SQLite dump named gospel-library.db in the project root to operate entirely offline. The same tools work unchanged.

Connecting a Client

There are two ways to connect to a client: locally and remotely. For developer use, local is recommended. Remote is easier to set up and uses the Cloudflare endpoint.

Local

Example config snippet:

{
  "mcpServers": {
    "gospel-library": {
      "command": "node",
      "args": ["/Users/path/to/repository/dist/stdio-server.js"],
      "env": { "GOSPEL_DB_PATH": "/Users/path/to/database/gospel-library.db"}
    }
  }
}

Remote

Option 1: One-Click Deploy (Recommended)

This will:

1. Fork the repository to your GitHub account
2. Set up a Cloudflare Workers deployment
3. Configure the D1 database automatically
4. Deploy your MCP server instance

Once deployed, use your Worker URL with the /sse endpoint in your MCP client.

Option 2: Manual Deployment

1. Clone this repository
2. Install dependencies: npm install
3. Set up Cloudflare Workers CLI: npm install -g wrangler
4. Login to Cloudflare: wrangler login
5. Create D1 database: wrangler d1 create gospel-library
6. Update database ID in wrangler.jsonc
7. Deploy: wrangler deploy

Option 3: Use Existing Deployment

No authentication is required to use this tool. You will need a client that supports remote MCP servers, such as Claude on a Pro plan or ChatGPT on a Pro plan.

URL: https://gospel-library-mcp.harrisadam42103.workers.dev/sse

Future Enhancements (Ideas)

• Ranking improvements or FTS.
• Citation generation referencing talk paragraphs.
• Caching of frequent queries.
• Footnotes
• Optionally allow for web scraping
• More resources from Gospel Library