---
name: epub
description: Read and extract content from EPUB ebook files. Use this skill when the user wants to read an EPUB file, extract text from an ebook, view EPUB metadata (title, author), list chapters or table of contents, search within EPUB content, or analyze ebook content.
---

# EPUB Reader Skill

Read EPUB ebook files and extract content as clean Markdown.

## Instructions

Use the epub-reader CLI tool to interact with EPUB files. The tool is located at:
`~/.claude/skills/epub/scripts/epub-reader/dist/index.js`

### Available Commands

#### 1. View Metadata
Get book information (title, author, publisher, date, description).

```bash
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js metadata "<path-to-epub>"
```

#### 2. List Table of Contents
View all chapters and their structure. Each entry shows `[ch: N]` indicating the chapter number to use with the `chapter` command.

```bash
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js toc "<path-to-epub>"
```

#### 3. Read Specific Chapter
Read a single chapter by number (1-indexed).

```bash
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js chapter "<path-to-epub>" <chapter-number>
```

#### 4. Read Entire Book
Extract the complete book as Markdown.

```bash
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js full "<path-to-epub>"
```

#### 5. Search Text
Find text occurrences with surrounding context.

```bash
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js search "<path-to-epub>" "<search-query>"
```

## Recommended Workflow

1. **Start with metadata** to understand what book you're working with
2. **View the TOC** to see available chapters and structure
3. **Read specific chapters** for targeted analysis, or use **full** for complete extraction
4. **Use search** to find specific topics, quotes, or references

## Open-Ended Search

For broad or conceptual queries like "what are the main themes in this book?" or "find all references to the protagonist's childhood", use **query expansion**:

1. **Expand the query** into multiple specific search terms using domain knowledge
   - Example: "protagonist's childhood" → search for character name, "young", "childhood", "memory", "father", "mother", "grew up", etc.
2. **Run searches in parallel** for each expanded term
3. **Synthesize results** by deduplicating and consolidating findings across searches

This approach leverages Claude's domain knowledge to catch synonyms, related concepts, and terminology variations that a simple keyword search would miss.

### Example

User asks: "What does the book say about the author's research methodology?"

Expand to searches:
- "methodology"
- "research"
- "study"
- "analysis"
- "data"
- "findings"
- "evidence"

Then consolidate the results into a comprehensive answer.

## Output Format

All output is clean Markdown:
- Headings preserved as `#`, `##`, etc.
- Lists, links, and emphasis converted properly
- Excessive whitespace cleaned up
- Chapter separators included for full extraction

## Examples

```bash
# What book is this?
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js metadata "/path/to/book.epub"

# Show me the chapters
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js toc "/path/to/book.epub"

# Read chapter 3
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js chapter "/path/to/book.epub" 3

# Find all mentions of "democracy"
node ~/.claude/skills/epub/scripts/epub-reader/dist/index.js search "/path/to/book.epub" "democracy"
```

## Notes

- Chapter numbers are 1-indexed (first chapter is 1, not 0)
- Use the `[ch: N]` reference from the TOC output to find the correct chapter number
- Paths with spaces must be quoted
- Large books may produce substantial output with the `full` command
- Search results show up to 5 matches per chapter with context