kindle-mcp
Transforms Kindle highlight exports into structured personal memory summaries saved to Notion. Parses HTML and text files to generate book insights including themes, key ideas, and actionable takeaways using the host model's context window without external AI API calls.
README
kindle-mcp
An MCP (Model Context Protocol) server that turns your Kindle highlights into structured personal book memory summaries saved directly to Notion.
Zero external AI API calls. All reasoning is performed by the host model (Claude Desktop) in its own context window. The server parses files, builds prompts, and writes to Notion — nothing else.
What it does
- You share a Kindle highlight export file in Claude Desktop
- Claude parses your highlights, generates a personal memory summary, and saves it to your Notion database
- You get back a Notion page URL — nothing else shown in chat
All automatic. No copy-pasting. No manual steps.
Requirements
- Claude Desktop
- Notion account
- Node.js 18+ installed
- Notion MCP server running alongside this one (for first-run auto-setup)
Installation
1. Get a Notion API key
- Go to https://www.notion.so/my-integrations
- Click New integration → give it a name → click Submit
- Copy the Internal Integration Secret (starts with
secret_...)
2. Add to Claude Desktop config
Open your Claude Desktop config file:
- Mac:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add the following inside "mcpServers":
"kindle-mcp": {
"command": "npx",
"args": ["-y", "kindle-mcp"],
"env": {
"NOTION_API_KEY": "secret_..."
}
}
3. Restart Claude Desktop
Fully quit (Cmd+Q on Mac) and reopen. You should see the tools available via the hammer icon in the chat input.
First-time use
On first run, the server needs a Notion page to create the Kindle Book Summaries database under. If you also have the Notion MCP running, this happens automatically — Claude will create the page and database without asking you anything.
If you don't have the Notion MCP, create any page in Notion and connect your integration to it first:
- Open a Notion page → click
...top right → Connections → select your integration
After the first run, the database ID is saved locally at ~/.kindle-mcp/config.json and never needs to be set up again.
Usage
In Claude Desktop:
- Attach your Kindle export file (
.htmlfrom the Kindle app, orMy Clippings.txtfrom a Kindle device) - Say: "Use the process_kindle_export tool on this file"
- Claude processes everything and returns your Notion page URL
Supported Kindle export formats
| Format | How to get it |
|---|---|
| HTML | Kindle app (iOS/Android/Mac) → open book → Notes → Export |
| Plain text | My Clippings.txt on Kindle device via USB |
Both formats are auto-detected — no configuration needed.
Summary output structure
Each book gets a Notion page with:
| Field | Description |
|---|---|
personal_thesis |
One sentence capturing the core insight from your highlights |
core_themes |
3–5 recurring themes |
key_ideas |
5–8 distinct ideas drawn from your highlights |
actionable_takeaways |
3–5 concrete actions implied by your highlights |
reflection_questions |
3–5 questions your highlights raise |
memory_capsule |
A 3–4 sentence personal narrative distilling everything |
Summaries are generated only from your highlights — the model uses no external knowledge about the book.
Tools
| Tool | Description |
|---|---|
process_kindle_export |
Primary tool. Parses a Kindle file and orchestrates the full flow |
initialize_notion_database |
Creates the Notion database on first run (called automatically) |
push_to_notion |
Pushes a generated summary to Notion as a structured page |
parse_kindle_clippings |
Parses a Kindle file and returns raw highlights grouped by book |
generate_personal_summary |
Builds a prompt package for the host model to generate a summary |
Notion database schema
The Kindle Book Summaries database is created automatically with:
| Property | Type |
|---|---|
Name |
Title |
Author |
Rich Text |
Title |
Rich Text |
All summary content (thesis, themes, ideas, etc.) is written as page body blocks.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。