zlib-mcp
An MCP server that lets AI agents search and download books from z-library using the user's own account credentials.
README
zlib-mcp
A stdio MCP server that gives any AI agent tool — Claude Code, Codex CLI, Cursor, Claude Desktop — the ability to search z-library and download books.
Bring your own account. There is no shared backend, no API key, no proxy: the server runs on your machine, talks straight to z-library, and uses your credentials and your quota.
Tools
| Tool | What it does | Needs credentials |
|---|---|---|
zlib_search |
Search by title / author / ISBN, with format, language and year filters | yes |
zlib_get_download_url |
Get a direct download link for one book (no file written) | yes |
zlib_download |
Download a book into a directory you configured | yes |
zlib_limits |
Check today's remaining download allowance | yes |
zlib_login |
One-time helper: exchange email + password for remix credentials | no |
zlib_download only appears once you set ZLIB_DOWNLOAD_DIR — an MCP server that can write files anywhere by default is not an acceptable default, so you have to name the directory yourself.
Requirements
- Node.js ≥ 20
- A z-library account
Setup in 5 minutes
1. Get your credentials
If you already know your remix_userid / remix_userkey, skip ahead. Otherwise add the server with just your email and password (see the config snippets below), then ask your agent to run zlib_login once and put the returned remix_id / remix_key into the config permanently.
You can also run it straight from a terminal:
ZLIB_EMAIL=you@example.com ZLIB_PASSWORD='…' npx zlib-mcp
2. Add the server to your client
Every client takes the same three things: the command npx, the argument zlib-mcp, and an env block.
<details open> <summary><b>Claude Code</b></summary>
claude mcp add zlib \
--env ZLIB_REMIX_ID=123456 \
--env ZLIB_REMIX_KEY=your_remix_userkey \
--env ZLIB_DOWNLOAD_DIR="$HOME/Downloads/books" \
-- npx -y zlib-mcp
Or edit ~/.claude.json / .mcp.json directly using the JSON below.
</details>
<details> <summary><b>Claude Desktop</b> — <code>claude_desktop_config.json</code></summary>
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"zlib": {
"command": "npx",
"args": ["-y", "zlib-mcp"],
"env": {
"ZLIB_REMIX_ID": "123456",
"ZLIB_REMIX_KEY": "your_remix_userkey",
"ZLIB_DOWNLOAD_DIR": "/Users/you/Downloads/books"
}
}
}
}
</details>
<details> <summary><b>Cursor</b> — <code>.cursor/mcp.json</code> or <code>~/.cursor/mcp.json</code></summary>
{
"mcpServers": {
"zlib": {
"command": "npx",
"args": ["-y", "zlib-mcp"],
"env": {
"ZLIB_REMIX_ID": "123456",
"ZLIB_REMIX_KEY": "your_remix_userkey",
"ZLIB_DOWNLOAD_DIR": "/Users/you/Downloads/books"
}
}
}
}
</details>
<details> <summary><b>Codex CLI</b> — <code>~/.codex/config.toml</code></summary>
[mcp_servers.zlib]
command = "npx"
args = ["-y", "zlib-mcp"]
[mcp_servers.zlib.env]
ZLIB_REMIX_ID = "123456"
ZLIB_REMIX_KEY = "your_remix_userkey"
ZLIB_DOWNLOAD_DIR = "/Users/you/Downloads/books"
</details>
3. Try it
Find me Kleppmann's Designing Data-Intensive Applications in epub, then download the first result.
Configuration
| Variable | Required | Default | Notes |
|---|---|---|---|
ZLIB_REMIX_ID |
one of two | — | Your remix_userid |
ZLIB_REMIX_KEY |
one of two | — | Your remix_userkey |
ZLIB_EMAIL |
one of two | — | Fallback: exchanged for remix credentials on first use |
ZLIB_PASSWORD |
one of two | — | Fallback, used with ZLIB_EMAIL |
ZLIB_HOST |
no | pkuedu.xyz |
Upstream mirror; change it if you get blocked |
ZLIB_DOWNLOAD_DIR |
no | (unset → zlib_download disabled) |
Where downloads are written |
ZLIB_MAX_DOWNLOAD_BYTES |
no | 524288000 (500 MB) |
Files above this need allow_large: true |
ZLIB_TIMEOUT_MS |
no | 20000 |
Per-request connect timeout |
ZLIB_CREDENTIAL_CACHE |
no | 1 |
Set 0 to never write ~/.zlib-mcp/credentials.json |
ZLIB_LOG_LEVEL |
no | info |
debug / info / warn / error / silent; all logs go to stderr |
Credential precedence
ZLIB_REMIX_ID+ZLIB_REMIX_KEY- Cached credentials from a previous
ZLIB_EMAILlogin (~/.zlib-mcp/credentials.json, mode600) ZLIB_EMAIL+ZLIB_PASSWORD→ logged in on first tool call, not at startup
The cache exists so a client restart does not trigger a fresh login every time — repeated logins are what makes z-library's anti-abuse system notice you. It stores only the remix id and key; your password is never written to disk, logged, or returned by any tool. On Windows the 600 mode is a no-op (the OS ignores POSIX permissions) — set ZLIB_CREDENTIAL_CACHE=0 if that matters to you.
If nothing is configured the server still starts and lists its tools; calling one returns instructions on what to set. It does not crash — a crashed MCP server just shows up as "unavailable" in most clients, with nothing to debug.
Troubleshooting
"Upstream host … appears to be blocking this request" — the mirror is behind an anti-bot wall. Set ZLIB_HOST to another one and restart the client. Known mirrors change often; 1lib.sk is currently blocked, pkuedu.xyz currently works. Anything that serves the same /eapi/* endpoints will do.
"z-library rejected the current credentials" — your remix key expired. Run zlib_login again and update the config. If you use the email/password fallback, delete ~/.zlib-mcp/credentials.json to force a fresh login.
"download quota reached" — free accounts get a small number of downloads per day. zlib_limits shows the counter; it resets on z-library's side at midnight UTC.
Nothing appears in the client — check the client's MCP log; this server writes all diagnostics to stderr. ZLIB_LOG_LEVEL=debug makes it chattier.
zlib_download is missing — you did not set ZLIB_DOWNLOAD_DIR. That is by design.
Development
pnpm install
pnpm check # format check → lint → typecheck → tests
pnpm build
To try an unreleased version straight from git, point your client's command/args at
npx / ["-y", "github:shiyi-0x7f/zlib-mcp"] — the prepare script builds it on install.
Tests never hit the real upstream — fetch is stubbed everywhere.
Legal
This tool only provides API access to your own z-library account. It hosts nothing, distributes nothing, and ships no copyrighted content. Making sure your use of it is lawful where you are is on you. Your account, your quota, your risk — an account banned for abuse is yours to lose.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.