OneNote MCP Server
Enables Claude to read and write to local Microsoft OneNote notebooks by parsing backup files and utilizing the OneNote COM API. It allows for searching notes, creating pages, and appending content without requiring Azure registration or API keys.
README
OneNote MCP Server
An MCP (Model Context Protocol) server that gives Claude access to your local Microsoft OneNote notebooks. It reads .one files directly from disk and writes to OneNote via the COM API — no Azure registration, no API keys, no authentication required.
What It Does
This server parses the OneNote backup files that the desktop app stores locally and exposes them as tools that Claude can use to browse, read, and write to your notes.
Reading Tools
| Tool | Description |
|---|---|
list_notebooks |
List all locally available OneNote notebooks (from backup files) |
list_sections |
List all sections in a specific notebook |
read_section |
Read the full text content of a section |
search_notes |
Search for text across all notebooks and sections |
list_all_sections |
List every section across every notebook |
get_notebook_summary |
Get a notebook overview with content previews |
Writing Tools
| Tool | Description |
|---|---|
list_live_notebooks |
List notebooks/sections from the running OneNote app |
create_page |
Create a new page in any notebook section |
list_live_pages |
List pages in a section (with IDs for appending) |
append_to_page |
Append content to an existing page |
Writing tools use the OneNote COM API and require the OneNote desktop app to be running on Windows.
Prerequisites
- Python 3.12+
- uv (recommended) or pip
- Microsoft OneNote desktop app (with local backup files)
Installation
git clone https://github.com/mhzarem/onenote-mcp.git
cd onenote-mcp
uv sync
Or with pip:
git clone https://github.com/mhzarem/onenote-mcp.git
cd onenote-mcp
pip install "mcp[cli]" pyOneNote
Setup
Claude Code
claude mcp add --transport stdio onenote -- uv --directory /path/to/onenote-mcp run server.py
Verify it's connected:
claude mcp list
Claude Desktop
Add this to your claude_desktop_config.json:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"onenote": {
"command": "uv",
"args": [
"--directory",
"/absolute/path/to/onenote-mcp",
"run",
"server.py"
]
}
}
}
On Windows, use the full path to uv.exe and double backslashes:
{
"mcpServers": {
"onenote": {
"command": "C:\\Users\\YOUR_USER\\.local\\bin\\uv.exe",
"args": [
"--directory",
"C:\\path\\to\\onenote-mcp",
"run",
"server.py"
]
}
}
}
Restart Claude Desktop after saving.
Where It Reads Files From
By default, the server reads from the OneNote desktop app's local backup directory:
C:\Users\<user>\AppData\Local\Microsoft\OneNote\16.0\Backup\
To use a different location, set the ONENOTE_BACKUP_DIR environment variable:
# Claude Code
claude mcp add --transport stdio --env ONENOTE_BACKUP_DIR=/path/to/notes onenote -- uv --directory /path/to/onenote-mcp run server.py
# Or export it
export ONENOTE_BACKUP_DIR=/path/to/your/onenote/files
Usage Examples
Once connected, you can ask Claude:
Reading:
- "List my OneNote notebooks"
- "Show me the sections in my Machine Learning notebook"
- "Read my Algorithm notes"
- "Search my notes for transformers"
- "Give me a summary of my Programming notebook"
Writing:
- "Create a new page in My Notebook / Quick Notes titled 'Meeting Notes'"
- "Add my interview prep notes to OneNote"
- "Append today's summary to my existing page"
How It Works
Reading:
- Scans the OneNote backup directory for
.onefiles - Organizes them by notebook and section (grouping backup versions together)
- Uses pyOneNote to parse the binary
.oneformat - Extracts
RichEditTextUnicodetext content from each section
Writing:
- Connects to the running OneNote desktop app via the COM API
- Uses PowerShell subprocess calls to create pages and update content
- Supports HTML formatting in page content
Limitations
- Reading: Uses OneNote desktop backup files — not OneDrive-only notebooks without local backup
- Reading: Extracts text content only; images and embedded files are not included
- Writing: Requires Windows with the OneNote desktop app installed
- Writing: The OneNote app must be installed (it doesn't need to be open — the COM API will start it)
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。