Read-only search of books and authors on Open Library

Who it's for

For people who want an assistant to search book and author information on Open Library. Official docs describe it as an MCP server for the Open Library API, used to search for book and author information. The repository title also says it talks to the Internet Archive's Open Library API.

This is not a local holdings directory and not a lending system. Every named tool is a read-only lookup. No API key field appears in the official config.

Official page:

https://github.com/8enSmith/mcp-open-library

First try in 3 minutes

Official docs say Nothing to install or build. Point the client at this package; the first run is pulled by npx:

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

From v1.0.3 it is published to the official MCP Registry as io.github.8enSmith/mcp-open-library; clients that support the registry can install by that name. Do not hunt for an installer that never appears.

Prerequisites: this machine can run npx. The first try should use a book title or author name you already know; do not invent an ISBN or Author Key. Official docs do not require a secret.

Config location: mcpServers.mcp-open-library. Official docs do not name a required client product or a config-file path.

The first call must use the named tool search_books. It can search by any combination of query, title, author, subject, place, person, publisher, and ISBN. Success: one page of results comes back with num_found; default limit is 10, at most 50. Do not first call get_author_info, which needs an Author Key.

readOnlyHint is not zero risk. Official docs also write openWorldHint: requests leave this machine and hit the Open Library API. The client may skip a “this will change things” confirm box, but the network and privacy boundary remain.

Use cases

Case 1: find books by combined conditions. Call search_books. Results are paged: each page has at most limit items (default 10, maximum 50), turn pages with offset (maximum 1000), and num_found is returned. Official docs do not describe sort or language filters.

Case 2: a narrower lookup by book title or author name. Call get_book_by_title, or call get_authors_by_name. Official docs only say search by title / name; they do not write fuzzy-match rules. If you already have an Open Library Author Key, you can use get_author_info for detail; without a Key, do not invent one.

Limits

Capability limit: tools named in the consecutive notes are search_books, get_book_by_title, get_authors_by_name, get_author_info. Do not invent cover or ID-fetch tools that never appear.

Paging limit: limit defaults to 10, maximum 50; offset maximum 1000. Do not assume paging beyond the official numbers.

Install limit: pull with npx, or the registry name io.github.8enSmith/mcp-open-library. Official docs do not describe an offline cache.