find-a-book-mcp

An MCP server for finding and downloading books from LibGen and Z-Library. Works with Claude Code, Claude Desktop, and any MCP-compatible AI agent.

| Source | Tools | Account needed? | |--------|-------|----------------| | Open Library | lookup_metadata | No | | Library Genesis | get_formats, search_books, download_book | No | | Z-Library | search_zlibrary, download_zlibrary | Yes (free) |

Quick rule of thumb: Start with search_books (LibGen, no login). If the book isn't there or you want a Chinese novel / recent release, switch to search_zlibrary (Z-Library).

Installation

Claude Desktop / Claude Code

Add to settings.json (no install needed — npx handles it automatically):

{
  "mcpServers": {
    "find-a-book": {
      "command": "npx",
      "args": ["-y", "find-a-book-mcp"],
      "env": {
        "ZLIBRARY_EMAIL": "[email protected]",
        "ZLIBRARY_PASSWORD": "yourpassword"
      }
    }
  }
}

Omit the env block if you only need LibGen (no Z-Library account required).

Manual

git clone https://github.com/Xiaochengzi2048/find-a-book-mcp
cd find-a-book-mcp
npm install
node server.js

Z-Library Setup

Z-Library has wider coverage than LibGen — especially for novels, Chinese books, and recent titles. Requires a free account at z-library.sk.

Option A — Email + password (auto-login on first use):

export [email protected]
export ZLIBRARY_PASSWORD=yourpassword

Option B — Token (skip login, slightly faster):

export ZLIBRARY_REMIX_USERID=12345678
export ZLIBRARY_REMIX_USERKEY=abcdef1234567890...

To get your token: log in at z-library.sk, open DevTools → Application → Cookies → copy remix_userid and remix_userkey.

Tools

lookup_metadata — Verify book info before searching

Looks up authoritative metadata from Open Library (free, no login). Use this before searching LibGen when you're not sure about the exact title or author — LibGen's scraped titles are often noisy ("Summary of…", alternate editions, etc.).

Parameters:

• query (string, optional) — Title, author, or keywords
• isbn (string, optional) — ISBN-10 or ISBN-13 for exact lookup (takes precedence over query)

Returns: Up to 5 candidates: { title, author, year, publisher, isbn, cover, openlibrary }. ISBN lookups also include pages.

When to use: You have a rough title/author and want to confirm the canonical version before searching. Or you have an ISBN and want clean metadata.

search_books — Search LibGen

Search by title, author, or keywords. Returns 5 results per page with pagination.

Parameters:

• query (string, required) — Title, author, or keywords
• page (number, optional, default 1) — Page number (5 results per page)
• extensions (string[], optional) — Format filter, e.g. ["epub", "pdf", "mobi"]

Returns:

{
  "books": [
    {
      "title": "The Psychology of Money",
      "author": "Morgan Housel",
      "year": "2020",
      "language": "English",
      "extension": "EPUB",
      "size": "920 kB",
      "size_bytes": 942080,
      "md5": "D7FF6458..."
    }
  ],
  "displayPage": 1,
  "totalDisplayPages": 4,
  "totalCount": 18,
  "hasMore": true,
  "hasPrev": false
}

When to use: Most non-fiction, English books, older titles. No account needed.

get_formats — See all available formats for a book

Same search as search_books, but deduplicates by title + author and groups all formats together. Fetches up to 75 results in parallel.

Parameters:

• query (string, required) — Title, author, or keywords

Returns:

[
  {
    "title": "The Psychology of Money",
    "author": "Morgan Housel",
    "year": "2020",
    "language": "English",
    "formats": [
      { "extension": "EPUB", "size": "920 kB", "size_bytes": 942080, "md5": "D7FF6458..." },
      { "extension": "MOBI", "size": "1.1 MB", "size_bytes": 1153433, "md5": "A3CC9012..." },
      { "extension": "PDF",  "size": "3.2 MB", "size_bytes": 3355443, "md5": "B9DE3F21..." }
    ]
  }
]

When to use: You want to pick a specific format (e.g. EPUB for e-reader, PDF for reference), or you want to see everything available in one call.

download_book — Download from LibGen

Download a book by its MD5 hash. Automatically tries multiple mirrors in parallel: libgen.li → libgen.rs → libgen.st → library.lol.

Parameters:

• md5 (string, required) — MD5 from search_books or get_formats results
• title (string, optional) — Used for the filename
• dest_dir (string, optional) — Save directory (defaults to system temp dir; created if missing)

Returns: { file_path, filename, size_bytes, extension }

search_zlibrary — Search Z-Library

Search Z-Library's catalog. Better than LibGen for novels, Chinese books, and recently published titles.

Requires: ZLIBRARY_EMAIL + ZLIBRARY_PASSWORD (or token env vars).

Parameters:

• query (string, required) — Title, author, or keywords
• page (number, optional, default 1) — Page number (10 results per page)
• extensions (string[], optional) — Format filter, e.g. ["epub", "pdf"]
• languages (string, optional) — Language filter, e.g. "chinese" or "english"
• year_from / year_to (number, optional) — Publication year range

Returns: Book list with zlibrary_id and zlibrary_hash (needed for download_zlibrary), plus title, author, year, language, format, and size.

When to use: LibGen doesn't have the book; it's a Chinese novel or recent release; you want to filter by language.

download_zlibrary — Download from Z-Library

Download a book using the zlibrary_id and zlibrary_hash from search_zlibrary results.

Requires: ZLIBRARY_EMAIL + ZLIBRARY_PASSWORD (or token env vars).

Free Z-Library accounts have a 10 downloads/day limit.

Parameters:

• zlibrary_id (string, required) — From search_zlibrary results
• zlibrary_hash (string, required) — From search_zlibrary results
• title (string, optional) — Used for the filename
• dest_dir (string, optional) — Save directory

Returns: { file_path, filename, size_bytes, extension }

Recommended Workflow

1. lookup_metadata("book title")        ← confirm exact title/author (optional but recommended)
2a. search_books("exact title")         ← try LibGen first (no login)
    → download_book(md5)
2b. search_zlibrary("exact title")      ← if not on LibGen, or want Chinese/recent
    → download_zlibrary(id, hash)

Tips for better results:

• Use lookup_metadata first when the title might be ambiguous — it normalizes spelling and finds the right edition
• Use get_formats to pick EPUB over PDF when both exist (EPUB is better on e-readers)
• Use languages: "chinese" in search_zlibrary to filter out English translations of Chinese books
• If search_books returns noisy results (summaries, study guides), add the author name to narrow it down

Environment Variables

| Variable | Description | |----------|-------------| | ZLIBRARY_EMAIL | Z-Library account email | | ZLIBRARY_PASSWORD | Z-Library account password | | ZLIBRARY_REMIX_USERID | Token auth: user ID from cookie (alternative to email/password) | | ZLIBRARY_REMIX_USERKEY | Token auth: user key from cookie | | LIBGEN_MIRRORS | Override LibGen mirror list (comma-separated URLs) when default domains rotate, e.g. https://libgen.is,https://libgen.gs |

Notes

• Downloaded files are saved to the system temp directory (/tmp/) by default — move them as needed
• size_bytes lets you decide delivery method (direct send vs. link for large files)
• Mirror failures are logged to stderr
• Supports EPUB, MOBI, PDF, AZW3, FB2, DJVU, and more

find-a-book-mcp（中文说明）

从 LibGen 和 Z-Library 搜索、下载电子书的 MCP 服务，兼容 Claude Code、Claude Desktop 及所有支持 MCP 协议的 AI Agent。

| 数据源 | 工具 | 需要账号？ | |--------|------|-----------| | Open Library | lookup_metadata | 否 | | Library Genesis | get_formats, search_books, download_book | 否 | | Z-Library | search_zlibrary, download_zlibrary | 是（免费） |

简单判断原则： 先用 search_books（LibGen，无需登录）。找不到，或要搜中文小说、近期出版物，再切 search_zlibrary（Z-Library）。

安装

Claude Desktop / Claude Code

在 settings.json 中添加（无需手动安装，npx 自动处理）：

{
  "mcpServers": {
    "find-a-book": {
      "command": "npx",
      "args": ["-y", "find-a-book-mcp"],
      "env": {
        "ZLIBRARY_EMAIL": "你的邮箱",
        "ZLIBRARY_PASSWORD": "你的密码"
      }
    }
  }
}

只用 LibGen 的话可以省略 env 部分，无需账号。

手动安装

git clone https://github.com/Xiaochengzi2048/find-a-book-mcp
cd find-a-book-mcp
npm install
node server.js

Z-Library 配置

Z-Library 覆盖面比 LibGen 更广，尤其适合小说、中文书、近期出版物。需要在 z-library.sk 注册免费账号。

方式 A — 邮箱 + 密码（首次使用时自动登录）：

export [email protected]
export ZLIBRARY_PASSWORD=yourpassword

方式 B — Token（跳过登录，稍快）：

export ZLIBRARY_REMIX_USERID=12345678
export ZLIBRARY_REMIX_USERKEY=abcdef1234567890...

获取方式：在 z-library.sk 登录后，打开 DevTools → Application → Cookies，复制 remix_userid 和 remix_userkey。

工具说明

lookup_metadata — 搜索前校验书目信息

从 Open Library 查询权威元数据（免费，无需账号）。在搜 LibGen 之前用它确认准确的书名/作者——LibGen 抓取的书名经常有噪音（"Summary of…"、不同版次等）。

参数：

• query（可选）— 书名、作者或关键词
• isbn（可选）— ISBN-10 或 ISBN-13，精确查询（优先级高于 query）

返回： 最多 5 个候选：{ title, author, year, publisher, isbn, cover, openlibrary }。ISBN 查询另含 pages。

适用场景： 书名不确定时先校验规范写法；有 ISBN 时直接拿干净元数据。

search_books — LibGen 搜索

按书名、作者或关键词搜索，每页返回 5 条，支持翻页。

参数：

• query（必填）— 书名、作者或关键词
• page（可选，默认 1）— 页码，每页 5 条
• extensions（可选）— 格式过滤，如 ["epub", "pdf", "mobi"]

返回字段： 书名、作者、出版社、年份、语言、格式、大小（含 size_bytes）、MD5，以及翻页信息（hasMore / hasPrev / totalCount）。

适用场景： 大多数非虚构类书籍、英文书、早期出版物。无需账号。

get_formats — 查看所有可用格式

与 search_books 搜索逻辑相同，但将同一书名+作者的所有格式归并展示。并行抓取最多 75 条结果，覆盖更全面。

参数：

• query（必填）— 书名、作者或关键词

返回示例：

[
  {
    "title": "置身事内",
    "author": "兰小欢",
    "year": "2021",
    "language": "Chinese",
    "formats": [
      { "extension": "EPUB", "size": "2.3 MB", "size_bytes": 2411724, "md5": "..." },
      { "extension": "PDF",  "size": "8.1 MB", "size_bytes": 8493466, "md5": "..." }
    ]
  }
]

适用场景： 想选特定格式（如 EPUB 用于阅读器，PDF 用于查阅），或想一次看全所有可用格式。

download_book — LibGen 下载

通过 MD5 下载书籍，自动并行尝试多个镜像：libgen.li → libgen.rs → libgen.st → library.lol。

参数：

• md5（必填）— 来自 search_books 或 get_formats 的 MD5
• title（可选）— 书名，用于文件命名
• dest_dir（可选）— 保存目录（默认系统临时目录，不存在会自动创建）

返回： { file_path, filename, size_bytes, extension }

search_zlibrary — Z-Library 搜索

在 Z-Library 搜索书籍，覆盖面比 LibGen 更广，尤其适合小说、中文书和近期出版物。

需要： ZLIBRARY_EMAIL + ZLIBRARY_PASSWORD（或 Token 环境变量）。

参数：

• query（必填）— 书名、作者或关键词
• page（可选，默认 1）— 页码，每页 10 条
• extensions（可选）— 格式过滤，如 ["epub", "pdf"]
• languages（可选）— 语言过滤，如 "chinese" 或 "english"
• year_from / year_to（可选）— 出版年份范围

返回： 书籍列表，含 zlibrary_id 和 zlibrary_hash（下载用），以及书名、作者、年份、语言、格式、大小。

适用场景： LibGen 没有该书；中文小说或近期出版物；需要按语言过滤。

download_zlibrary — Z-Library 下载

用 search_zlibrary 返回的 zlibrary_id + zlibrary_hash 下载书籍。

需要： ZLIBRARY_EMAIL + ZLIBRARY_PASSWORD（或 Token 环境变量）。

免费账号每天限下载 10 本。

参数：

• zlibrary_id（必填）— 来自 search_zlibrary 结果
• zlibrary_hash（必填）— 来自 search_zlibrary 结果
• title（可选）— 书名，用于文件命名
• dest_dir（可选）— 保存目录

返回： { file_path, filename, size_bytes, extension }

推荐使用流程

1. lookup_metadata("书名")              ← 确认准确书名/作者（可选但推荐）
2a. search_books("准确书名")            ← 先试 LibGen（无需登录）
    → download_book(md5)
2b. search_zlibrary("准确书名")         ← LibGen 没有，或要搜中文/近期出版物
    → download_zlibrary(id, hash)

用好这个工具的几个技巧：

• 书名有歧义时先用 lookup_metadata 校准——它能找到规范拼写和正确版次
• get_formats 可以选格式：有 EPUB 时优先选 EPUB（阅读器更友好）
• 搜中文书时加 languages: "chinese" 过滤掉英译本
• search_books 结果有噪音（摘要版、学习指南）时，加上作者名再搜

环境变量

| 变量名 | 说明 | |--------|------| | ZLIBRARY_EMAIL | Z-Library 账号邮箱 | | ZLIBRARY_PASSWORD | Z-Library 账号密码 | | ZLIBRARY_REMIX_USERID | Token 认证：Cookie 中的 userid（可替代邮箱/密码） | | ZLIBRARY_REMIX_USERKEY | Token 认证：Cookie 中的 userkey | | LIBGEN_MIRRORS | 覆盖默认 LibGen 镜像列表（逗号分隔的 URL），域名轮换时使用，如 https://libgen.is,https://libgen.gs |

注意事项

• 文件默认保存在系统临时目录（/tmp/），请自行移至目标位置
• size_bytes 字段方便判断发送方式（小文件直发，大文件给链接）
• 镜像失败时写入 stderr，便于排查
• 支持 EPUB、MOBI、PDF、AZW3、FB2、DJVU 等格式