← 返回 Skills 市场

Book Reader - Learn & Grow Every Day

Name: Book Reader - Learn & Grow Every Day
Author: josharsh

作者 josharsh · GitHub ↗ · v1.0.0

cross-platform ⚠ suspicious

1388

总下载

当前安装

版本数

在 OpenClaw 中安装

/install book-reader

功能描述

Read EPUB, PDF, or TXT books from multiple sources with progress tracking, smart chunking, and summary extraction for efficient learning.

使用说明 (SKILL.md)

book-reader

Read books (epub, pdf, txt) from various sources with progress tracking.

Purpose

Enable AI agents to read full-length books for learning, summarization, and knowledge extraction.

Features

Multiple sources: Anna's Archive, Project Gutenberg, local files
Format support: EPUB, PDF, TXT
Progress tracking: Remember where you left off
Smart chunking: Read books in digestible sections
Summary generation: Extract key insights as you read

Tools Required

curl or wget - Download books
pandoc - Convert EPUB to text (optional, fallback to python)
pdftotext (poppler-utils) - Extract PDF text
Python 3 with ebooklib and beautifulsoup4 (for EPUB parsing)

Usage

Search for a book

./book-reader.sh search "Thinking Fast and Slow"

Download a book

./book-reader.sh download \x3Cbook-id> [output-file]

Read a book (with progress tracking)

./book-reader.sh read \x3Cfile> [--from-page N] [--pages N]

Show reading progress

./book-reader.sh status

Installation

# Install dependencies
sudo apt-get install poppler-utils pandoc  # Linux
# brew install poppler pandoc  # macOS

pip3 install ebooklib beautifulsoup4 lxml

# Make executable
chmod +x book-reader.sh

Book Sources

Project Gutenberg (70k+ public domain books)
- API: https://gutendex.com
- Free, legal, no DRM
Anna's Archive (shadow library)
- Millions of books, papers, comics
- Legal gray area depending on jurisdiction
- Use responsibly
Local files (your own epub/pdf collection)

Reading State

Progress tracked in ~/.openclaw/workspace/memory/reading-state.json:

{
  "currentBook": "Thinking, Fast and Slow",
  "file": "/path/to/book.epub",
  "totalPages": 499,
  "pagesRead": 127,
  "lastRead": 1770957900,
  "bookmarks": [50, 200],
  "notes": "Interesting insight about System 1 vs System 2..."
}

Example Workflow

# Find the book
./book-reader.sh search "Daniel Kahneman Thinking"

# Download it
./book-reader.sh download 12345 ~/books/thinking-fast-slow.epub

# Start reading
./book-reader.sh read ~/books/thinking-fast-slow.epub --pages 50

# Continue later
./book-reader.sh read ~/books/thinking-fast-slow.epub --pages 50

# Check progress
./book-reader.sh status

Privacy & Ethics

Public domain books (Gutenberg): Fully legal
Copyrighted books: Check your local laws
Consider buying books you find valuable to support authors
Use for personal learning, not redistribution

Limitations

PDF OCR quality varies
DRM-protected books not supported (by design)
Large PDFs may be slow to parse
EPUB formatting may be imperfect in plain text

Use responsibly. Support authors when possible.

安全使用建议

This skill appears to do what it says: download books, extract text, and track reading progress in ~/.openclaw/workspace. Before installing or running: 1) Ensure you have the runtime tools the script actually needs — SKILL.md omits jq (the script uses jq for JSON parsing), and the Python EPUB path requires ebooklib and bs4/lxml if pandoc is not present. 2) Be mindful of legality and ethics: the script can download from arbitrary URLs and references Anna's Archive (a shadow library); avoid downloading copyrighted material unless you have the right to do so. 3) Review network usage policies: downloads come from Gutendex or user-provided URLs, so only use trusted sources. 4) The script writes reading-state.json in your workspace and may include file paths and notes — treat that as local data. If you need higher assurance, run the script in a sandbox or review the code locally before giving it files or network access.

功能分析

Type: OpenClaw Skill Name: book-reader Version: 1.0.0 The skill is classified as suspicious due to the `download_book` function in `book-reader.sh` allowing the agent to specify an arbitrary output file path for downloaded content. While the script itself does not exhibit malicious intent, this capability, if exploited via prompt injection against the AI agent, could lead to arbitrary file writes to sensitive locations (e.g., `/etc/cron.d/`, `~/.ssh/authorized_keys`), potentially enabling persistence, privilege escalation, or remote code execution. The `SKILL.md` and `README.md` also mention Anna's Archive, a 'shadow library' with legal gray areas, which, while not a code vulnerability, introduces an ethical/legal risk for the agent's operation.

能力评估

✓ Purpose & Capability

The script and SKILL.md align with the stated purpose: searching Gutenberg, downloading EPUB/PDF/TXT, extracting text, and storing reading progress. Declared dependencies (pandoc, pdftotext, Python ebook libs) map to the code's EPUB/PDF handling. One small mismatch: the script uses jq for JSON processing but jq is not mentioned in SKILL.md or the registry metadata.

✓ Instruction Scope

Runtime instructions and the script stay within the declared scope: downloading books (from user-supplied URLs or Gutendex), extracting text, and saving progress to a local workspace (~/.openclaw/workspace). The skill does not instruct reading unrelated system files or exfiltrating data to third-party endpoints beyond fetching book sources and calling the Gutendex API. It mentions Anna's Archive as a source but does not automate access to it (the script points users to the site manually).

✓ Install Mechanism

There is no install spec (instruction-only), so nothing arbitrary is downloaded or executed during installation. Dependencies are installed by the user via OS package manager or pip per SKILL.md. This is the lower-risk pattern.

✓ Credentials

The skill requests no secrets or privileged environment variables. It optionally respects OPENCLAW_WORKSPACE to change its storage location; otherwise it writes state under ~/.openclaw/workspace, which is proportional to its functionality.

✓ Persistence & Privilege

always is false and the skill does not request persistent platform-wide privileges. It stores only its own reading-state file under its workspace and does not modify other skills or system-wide agent settings.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install book-reader
安装完成后，直接呼叫该 Skill 的名称或使用 /book-reader 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Initial release: Read full books with progress tracking. 70K+ free books from Project Gutenberg. Built for AI agents on continuous learning journeys.

元数据

Slug book-reader

版本 1.0.0

许可证 —

累计安装 9

当前安装数 9

历史版本数 1

常见问题