/ Directory / Playground / ebook-mcp
← All servers ↗ GitHub
● Community onebirdrocks ⚡ Instant

ebook-mcp

by onebirdrocks · onebirdrocks/ebook-mcp

Point Claude at your EPUB and PDF library — get metadata, TOC, and chapter content in structured form so the model can quote and summarize precisely.

ebook-mcp wraps Python's common EPUB/PDF libs into MCP tools. Unlike generic 'read a file', it exposes table of contents and chapter boundaries, so Claude can fetch just the chapter you're asking about instead of dumping 500 pages into context.

▶ Watch demo 📦 Install 💡 Use Cases

Why use it

Key features

Live Demo

What it looks like in practice

ebook.replay ▶ ready
0/0

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json  · Windows: %APPDATA%\Claude\claude_desktop_config.json
{
  "mcpServers": {
    "ebook": {
      "command": "uvx",
      "args": [
        "ebook-mcp"
      ],
      "_inferred": true
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json
{
  "mcpServers": {
    "ebook": {
      "command": "uvx",
      "args": [
        "ebook-mcp"
      ],
      "_inferred": true
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit
{
  "mcpServers": {
    "ebook": {
      "command": "uvx",
      "args": [
        "ebook-mcp"
      ],
      "_inferred": true
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "ebook": {
      "command": "uvx",
      "args": [
        "ebook-mcp"
      ],
      "_inferred": true
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json
{
  "mcpServers": [
    {
      "name": "ebook",
      "command": "uvx",
      "args": [
        "ebook-mcp"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json
{
  "context_servers": {
    "ebook": {
      "command": {
        "path": "uvx",
        "args": [
          "ebook-mcp"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add ebook -- uvx ebook-mcp

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use ebook-mcp

Find and quote a passage you vaguely remember

👤 Readers / researchers ⏱ ~5 min beginner

When to use: 'There was something in that book about X...' — you need the exact quote and can't be bothered to flip pages.

Prerequisites
  • The book file on disk — EPUB or PDF
Flow
  1. Load TOC
    Show me the TOC of ~/Books/deep-work.epub. Which chapter likely discusses shallow work vs deep work?✓ Copied
    → TOC + best-guess chapter
  2. Pull the chapter
    Get the text of chapter 3. Find the passage about shallow-work-as-default.✓ Copied
    → Quoted passage

Outcome: Quote in hand, citable, in 2 minutes.

Pitfalls
  • Scanned PDFs have no text layer — Run OCR first

Generate a chapter-by-chapter summary of a technical book

👤 Engineers learning from a book on their TBR ⏱ ~30 min intermediate

When to use: You want to decide whether to actually read a book, or need a refresher after finishing.

Flow
  1. TOC first
    Load TOC of book.pdf. Show me chapter titles and approximate page ranges.✓ Copied
    → Outline
  2. Per-chapter summary
    For each chapter, load its content and produce a 3-bullet summary. Output as a single Markdown doc.✓ Copied
    → Complete book digest

Outcome: A 1-2 page book digest you can share or revisit.

Pitfalls
  • Huge books blow the token budget if you load all chapters at once — Process one chapter at a time; write intermediate summaries to disk
Combine with: filesystem

Audit your ebook library's metadata

👤 Calibre users, ebook hoarders ⏱ ~15 min beginner

When to use: Your library has 500 ebooks and you want to find ones with missing title/author metadata.

Flow
  1. Scan
    For each .epub under ~/Books, get metadata. List any where title or author is missing or contains 'Unknown'.✓ Copied
    → Incomplete-metadata list

Outcome: Quick metadata hygiene pass.

Combine with: filesystem

Combinations

Pair with other MCPs for X10 leverage

ebook + local-rag

Build a searchable library index

For each book, get_chapter by chapter, ingest into local-rag with book title as source.✓ Copied
ebook + notion

Auto-populate your reading journal

Take this just-finished book; summarize each chapter; create a Notion page in 'Books Read'.✓ Copied

Tools

What this MCP exposes

ToolInputsWhen to callCost
get_metadata path: str First pass on any book free
get_toc path: str Before pulling chapter content free
get_chapter path, chapter_id or id_range Load a specific chapter free
get_pages path, from_page, to_page (PDF only) Page-range extraction for PDFs free

Cost & Limits

What this costs to run

API quota
None
Tokens per call
Chapters 2k-30k tokens depending on length
Monetary
Free (Apache 2.0)
Tip
Always load TOC first. Then load chapters one at a time rather than the whole book.

Security

Permissions, secrets, blast radius

Credential storage: None — local files only
Data egress: Book content goes to your LLM provider

Troubleshooting

Common errors and fixes

Unable to parse PDF

Scanned/image-only PDF has no text layer. Run ocrmypdf first.

Verify: pdftotext input.pdf - | head
EPUB chapter ID not found

IDs come from the TOC exactly. Call get_toc first.

Large PDF makes get_chapter timeout

Use get_pages with a smaller page range; batch.

Alternatives

ebook-mcp vs others

AlternativeWhen to use it insteadTradeoff
Calibre + custom scriptYou want full Calibre integrationMore setup; no MCP wrapper
pdfplumber / pymupdf directlyYou build your own toolingBuild your own MCP wrapper

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills