zlib-download

name	zlib-download
description	Search and download books from Z-Library and Anna's Archive. Use when user wants to find, search, download, or look up books, papers, or ebooks. Trigger phrases include "find book", "search book", "download book", "找书", "下载书籍", "搜书", "book search", "zlibrary", "anna's archive".

Book Tools

Search and download books from multiple sources through a unified CLI.

Backends

Backend	Source	Auth Required	Best For
zlib	Z-Library (EAPI)	Email + Password	Largest catalog, direct download
annas	Anna's Archive	API Key (donation)	Aggregated sources, multiple mirrors

First-Time Setup

On first invocation, run the setup check and guide the user through configuration interactively.

Step 1: Check Dependencies

bash ${SKILL_PATH}/scripts/setup.sh check

Output is key=value pairs. Check each:

Key	OK	Missing Action
PYTHON	ok	Python 3 not found — user must install it
REQUESTS	ok	Run bash ${SKILL_PATH}/scripts/setup.sh install-deps
ANNAS_BINARY	ok	Run bash ${SKILL_PATH}/scripts/setup.sh install-annas (optional)

Step 2: Configure Credentials

Credentials are stored in ~/.claude/book-tools/.env. Create the file from the skill's bundled template:

mkdir -p ~/.claude/book-tools
cp ${SKILL_PATH}/scripts/.env.example ~/.claude/book-tools/.env

The .env file looks like this:

# Z-Library credentials
ZLIB_EMAIL=[e-mail]
ZLIB_PASSWORD=your_password_here

# Anna's Archive (optional, requires donation for API key)
# ANNAS_SECRET_KEY=your_api_key_here

IMPORTANT: Do NOT ask the user for credentials directly in chat. Instead:

1. Create the .env file (or .env.example template)
2. Tell the user to edit ~/.claude/book-tools/.env with their credentials
3. Wait for the user to confirm they've filled it in
4. Then proceed with search

Alternatively, credentials can be set via CLI (less recommended — visible in shell history):

python3 ${SKILL_PATH}/scripts/book.py config set --zlib-email "[e-mail]" --zlib-password "password"

Step 3: Verify

python3 ${SKILL_PATH}/scripts/book.py setup

Expected output when Z-Library is configured:

{
  "zlib": { "requests_installed": true, "configured": true },
  "annas": { "binary_found": true, "api_key_configured": false }
}

If configured is true, the skill is ready to use.

Credential Storage Details

Two sources are merged (.env values take priority):

Source	Path	Format
.env file	~/.claude/book-tools/.env	KEY=value per line
Config JSON	~/.claude/book-tools/config.json	JSON (auto-managed)

On first successful Z-Library login, remix tokens are cached in config.json — subsequent calls skip the email/password login and use tokens directly.

Workflow

The typical flow is: search → pick → download.

1. Search

# Auto-detect backend (tries zlib first, then annas)
python3 ${SKILL_PATH}/scripts/book.py search "machine learning" --limit 10

# Z-Library with filters
python3 ${SKILL_PATH}/scripts/book.py search "deep learning" --source zlib --lang english --ext pdf --limit 5

# Anna's Archive
python3 ${SKILL_PATH}/scripts/book.py search "reinforcement learning" --source annas

# Chinese books
python3 ${SKILL_PATH}/scripts/book.py search "莱姆 索拉里斯" --source zlib --lang chinese --limit 5

Output (JSON to stdout):

{
  "source": "zlib",
  "count": 5,
  "books": [
    {
      "source": "zlib",
      "id": "12345",
      "hash": "abc123def",
      "title": "Deep Learning",
      "author": "Ian Goodfellow",
      "year": "2016",
      "language": "english",
      "extension": "pdf",
      "filesize": "22.5 MB"
    }
  ]
}

2. Present Results to User

After searching, present results as a numbered table so the user can pick:

| # | Title | Author | Year | Format | Size |
|---|-------|--------|------|--------|------|
| 1 | Deep Learning | Ian Goodfellow | 2016 | pdf | 22.5 MB |
| 2 | ... | ... | ... | ... | ... |

If results span multiple languages or editions, group them by language or category with sub-headings for clarity.

Ask: "Which book would you like to download? (number)"

3. Download

# Z-Library download (needs id + hash from search results)
python3 ${SKILL_PATH}/scripts/book.py download --source zlib --id 12345 --hash abc123def -o ~/Downloads/

# Anna's Archive download (needs MD5 hash from search results)
python3 ${SKILL_PATH}/scripts/book.py download --source annas --hash a1b2c3d4e5 --filename "deep_learning.pdf" -o ~/Downloads/

Output:

{
  "source": "zlib",
  "status": "ok",
  "path": "~/Downloads/Deep Learning (Ian Goodfellow).pdf",
  "size": 23592960
}

4. Report to User

After download, report:

• File path (so user can open it)
• File size
• Any remaining download quota (Z-Library has daily limits)

Other Commands

Book Info (Z-Library only)

python3 ${SKILL_PATH}/scripts/book.py info --source zlib --id 12345 --hash abc123def

Returns full metadata: description, ISBN, pages, table of contents, etc.

Check Config

python3 ${SKILL_PATH}/scripts/book.py config show

Check Backend Status

python3 ${SKILL_PATH}/scripts/book.py setup

Error Handling

Error	Cause	Action
"Z-Library not configured"	No credentials	Guide user to edit ~/.claude/book-tools/.env
"Z-Library login failed"	Bad credentials or service down	Ask user to verify credentials. Z-Library domains change — if persistent, the vendored Zlibrary.py domain may need updating.
"annas-mcp binary not found"	Binary not installed	Run setup.sh install-annas
"Anna's Archive API key not configured"	No API key	Guide user to donate at Anna's Archive for API access, then add key to .env
Search timeout	Network issue	Retry once. If persistent, try the other backend.
"No backend available"	Neither backend configured	Walk through full setup flow from Step 1

Tips

• Z-Library has a daily download limit (usually 10/day for free accounts). Use info to check a book before downloading to avoid wasting quota.
• Anna's Archive requires an API key for both search and download (obtained via donation).
• For Chinese books, use --lang chinese with Z-Library for best results.
• If Z-Library is unreachable, automatically fall back to Anna's Archive with --source auto.
• When searching for a specific author in multiple languages, run parallel searches (e.g. English name + Chinese name) and merge results into one table.