MCP Open Library

What 8enSmith/mcp-open-library does

8enSmith/mcp-open-library connects MCP-compatible AI clients to the Open Library API for read-only book and author research. It returns selected JSON data rather than passing through complete API payloads. The available operations cover catalogue searches, author discovery, detailed records, and image URL lookup.

Book searches can use a general query or specific fields including title, author, subject, place, person, publisher, and ISBN. Title-only lookup is available separately. Book details can be retrieved with an ISBN, LCCN, OCLC identifier, or Open Library ID (OLID). Author records use an Open Library Author Key, while author photos require an author OLID.

How it works

Run the package as an MCP server over standard input/output. A client starts it with npx, and the package is fetched on first use. The repository also supports a manual clone, dependency installation, and build process.

Search tools return up to 10 results by default and include num_found, offset, and limit values for paging. The maximum page size is 50, and offsets can reach 1000. search_books requires at least one search criterion; its optional language and sort parameters can further refine results. It also accepts free-form Solr-style queries through q.

A search result represents a work and includes a best_edition object. Its edition_key is an OLID that can be supplied to get_book_by_id to retrieve the corresponding edition record. ISBN-10 and ISBN-13 values are included when Open Library has them.

All seven tools are read-only and advertise read-only and open-world hints. These annotations are advisory, so the MCP client determines whether confirmation is needed. API or argument failures are returned as tool results marked with isError.

Setup and configuration

The quick-start configuration launches the npm package directly:

{
  "mcpServers": {
    "mcp-open-library": {
      "command": "npx",
      "args": ["-y", "mcp-open-library"]
    }
  }
}

Add this entry to the configuration used by the MCP client and restart it. Claude Desktop uses claude_desktop_config.json; other clients use the equivalent MCP server configuration format.

For a local checkout, clone the repository, run npm install, and build with npm run build. The usage instructions use Node.js 22.23.4. The project can also be tested with the MCP Inspector after building: run npm run inspector, select STDIO, set the command to build/index.js, and connect.

Tools and capabilities

• search_books searches by combined catalogue fields and supports sorting and paging.
• get_book_by_title searches Open Library using a title.
• get_book_by_id retrieves a detailed book record from an ISBN, LCCN, OCLC, or OLID.
• get_authors_by_name searches for authors with paged results.
• get_author_info retrieves an author record from an Open Library Author Key.
• get_author_photo returns an author photo URL when a photo exists.
• get_book_cover returns a cover URL using ISBN, OCLC, LCCN, OLID, or another supported ID.

The photo and cover tools report when no image exists instead of returning a URL for a blank placeholder.