annotation-extractor 0.4.1

MCP server for extracting annotations and highlights from e-readers (Kobo, Kindle, Boox)

annotation-extractor

MCP server for extracting annotations, highlights, and reading progress from e-readers.

Supports Kobo, Kindle, and Boox.

Install

pip install annotation-extractor

As an Agent Skill

Install as a reusable skill for any AI coding agent via skills.sh:

npx skills add rkaushik29/annotation-extractor-mcp

Or from source:

git clone https://github.com/rkaushik29/annotation-extractor-mcp.git
cd annotation-extractor
pip install -e .

Usage

Connect your e-reader via USB, add the MCP server to your agent, and start asking questions about your reading.

Client	Setup
Claude Code	claude mcp add annotations -- annotation-extractor
Claude Desktop	Add to claude_desktop_config.json: {"mcpServers": {"annotations": {"command": "annotation-extractor"}}}
OpenClaw	openclaw mcp add annotations -- annotation-extractor
Gemini CLI	Add to ~/.gemini/settings.json: {"mcpServers": {"annotations": {"command": "annotation-extractor"}}}
Codex	codex --mcp-config mcp.json with mcp.json: {"mcpServers": {"annotations": {"command": "annotation-extractor"}}}

Try it out

Once the MCP is connected and your e-reader is plugged in, try asking your agent:

"What are my highlights from the last book I read?"

"Search my annotations for anything about consciousness"

"Show me all the notes I took while reading Neuromancer"

Environment variables

If auto-detection doesn't find your device, you can point to the data source directly:

KOBO_DB_PATH=/path/to/KoboReader.sqlite annotation-extractor
KINDLE_CLIPPINGS_PATH=/path/to/My\ Clippings.txt annotation-extractor
BOOX_EXPORT_PATH=/path/to/boox-export annotation-extractor

For handwritten note extraction and rendering, these optional variables are also supported:

KINDLE_SCRIBE_PATH=/path/to/kindle-mount annotation-extractor
KINDLE_SCRIBE_CONVERTER=calibre-debug annotation-extractor
BOOX_NOTE_RENDERER=onyx_render.py annotation-extractor

Supported e-readers

Reader	Data source
Kobo	.kobo/KoboReader.sqlite via USB
Kindle	My Clippings.txt via USB
Boox	Annotation export directory (one .txt file per book) via USB

Handwritten notes are supported for:

• Kindle Scribe notebooks via .notebooks/ (raw export, optional render via KFX Input)
• Boox handwritten exports (.pdf / .png) and local backup .zip files (optional render)

Tools

Tool	Description
list_books	List all books, optionally filtered to those with annotations
get_annotations	Get highlights and notes for a specific book (by title or ID)
search_annotations	Full-text search across all highlights and notes
get_reading_progress	Reading progress, time spent, and status for all books
get_book_details	Detailed metadata for a specific book
get_handwritten_notes	List handwritten note artifacts (Kindle Scribe / Boox)
export_handwritten_notes	Export handwritten note artifacts to a local directory

All tools accept optional backend_name (e.g., "kobo", "kindle", "boox") and db_path parameters. When omitted, the server auto-detects connected devices.

Adding a new backend

1. Create src/annotation_extractor/backends/yourreader.py implementing EReaderBackend
2. Add your backend to _ALL_BACKENDS in registry.py
3. No changes needed to server.py or models.py

Development

git clone https://github.com/rohitkaushik/annotation-extractor.git
cd annotation-extractor
pip install -e ".[dev]"
pytest

License

MIT