notion-mcp-server
Enables natural language interaction with Notion workspace, including search, page/database management, markdown conversion, and productivity shortcuts.
README
Notion MCP Server
🚀 Powerful Notion integration for Claude — read, create, search, and manage your entire Notion workspace through natural language.
✨ Features
- 🔍 Smart Search — Find pages and databases instantly
- 📄 Page Management — Create, read, update, archive, duplicate pages
- 🎨 Full Markdown Support — Seamless markdown ↔ Notion blocks conversion
- 🗂 Database Operations — Query with filters, sort, bulk add rows
- ⚡ Productivity Shortcuts — Quick notes, daily journal, task management
- 🧱 Block-Level Control — Read, update, delete individual blocks
- 💬 Comments Support — Add and read page comments
- 🎓 Learning Dashboard — Track AI/ML or any learning database progress
- 🔒 Privacy First — All pages private by default
- ⚡ Fast & Async — Built with FastMCP, schema caching, auto-retry
🎬 What You Can Do
Ask Claude things like:
- "Search my Notion for 'machine learning'"
- "Create a page called 'Project Plan' with a roadmap"
- "Add 20 AI/ML topics to my Topics database"
- "Show me all tasks marked Critical priority"
- "What's on my schedule today?"
- "Add a quick note: meeting was great"
- "Show my learning dashboard"
📦 Installation
Prerequisites
- Python 3.10 or higher
- A Notion account
- Claude Desktop, Claude Code, or any MCP-compatible client
Install with uv (recommended)
uv tool install notion-mcp-server
Install with pip
pip install notion-mcp-server
🔧 Setup (5 minutes)
Each user needs their own Notion integration token. Tokens are personal — they grant access only to the pages you explicitly share with your integration. No
.envfile is needed when usinguvx; your token goes directly in the Claude config JSON (Step 3).
Step 1: Create a Notion Integration
- Go to notion.so/my-integrations
- Click "+ New integration"
- Name it (e.g., "Claude MCP")
- Select your workspace
- Click Submit
- Copy the Internal Integration Secret (starts with
secret_orntn_)
Step 2: Share Pages with Integration
For each page or database you want Claude to access:
- Open the page in Notion
- Click "..." menu (top right)
- Click "Connections" → "Connect to"
- Select your integration
💡 Pro tip: Sharing a parent page automatically gives access to all sub-pages. Create one "Workspace Hub" page and share that.
Step 3: Configure Claude Desktop
Mac/Linux:
~/Library/Application\ Support/Claude/claude_desktop_config.json
Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add this configuration:
{
"mcpServers": {
"notion": {
"command": "uvx",
"args": ["notion-mcp-server"],
"env": {
"NOTION_TOKEN": "secret_your_token_here",
"DEFAULT_PRIVACY": "private",
"DEFAULT_PARENT_PAGE_ID": "optional_parent_page_id"
}
}
}
}
🗂 Alternative: Project-level Setup via .mcp.json
Instead of editing the global Claude Desktop config, you can enable the Notion MCP server per-project by adding a .mcp.json file to the root of any codebase. This is ideal for team projects or when you want the server scoped to a specific workspace.
Create .mcp.json in your project root:
{
"mcpServers": {
"notion": {
"command": "uvx",
"args": ["notion-mcp-server"],
"env": {
"NOTION_TOKEN": "secret_your_token_here",
"DEFAULT_PRIVACY": "private",
"DEFAULT_PARENT_PAGE_ID": "optional_parent_page_id"
}
}
}
}
💡 Tip: Add
.mcp.jsonto your.gitignoreif it contains yourNOTION_TOKEN, or use environment variable references and keep secrets in a local.envfile.
Supported by Claude Code, Cursor, Windsurf, and other MCP-compatible editors that respect project-level config files.
Step 4: Restart Claude Desktop
Fully quit (including system tray) and reopen.
Step 5: Test It!
Ask Claude:
"Check my Notion connection"
If you see ✅ Connected — you're all set!
🛠 Available Tools (22 Total)
🔍 Search & Connection
search_notion(query, filter_type, limit)— Search pages/databasescheck_connection()— Verify integration works
📄 Page Operations
get_page(page_id_or_url)— Read page as markdowncreate_page(title, parent_id, content, privacy)— Create new pageappend_to_page(page_id_or_url, content)— Append contentupdate_page_title(page_id_or_url, new_title)— Rename pagearchive_page(page_id_or_url)— Soft-delete pagerestore_page(page_id_or_url)— Restore archived pageduplicate_page(page_id_or_url, new_title)— Clone page
🗂 Database Operations
list_databases()— List all databasesget_database_schema(database_id_or_url)— See columns and typesquery_database(database_id_or_url, filters, sort_by, ...)— Query rowsadd_database_row(database_id_or_url, properties, content)— Add rowupdate_database_row(page_id_or_url, properties)— Update rowbulk_add_rows(database_id_or_url, rows)— Add many rows at once
🧱 Block Operations
get_block_children(block_id_or_url)— Read all blocksupdate_block(block_id_or_url, new_content)— Edit a blockdelete_block(block_id_or_url)— Delete a block
💬 Comments
add_comment(page_id_or_url, comment_text)— Add a commentget_comments(page_id_or_url)— Read comments
⚡ Productivity Shortcuts
quick_note(content, title)— Quick note to default parentdaily_journal(entry)— Append timestamped entry to journaladd_task(title, due_date, priority, status, tags)— Smart-add taskget_today_tasks()— Tasks due today or overduecomplete_task(task_name_or_id)— Mark task doneget_recent_pages(limit)— Recently edited pageslearning_dashboard()— Learning progress overview
🎨 Markdown Support
Full bidirectional conversion:
| Markdown | Notion Block |
|---|---|
# Heading 1 |
Heading 1 |
## Heading 2 |
Heading 2 |
### Heading 3 |
Heading 3 |
- bullet |
Bulleted list |
1. numbered |
Numbered list |
- [ ] todo |
To-do (unchecked) |
- [x] done |
To-do (checked) |
> quote |
Quote block |
```code``` |
Code block |
--- |
Divider |
**bold** |
Bold text |
*italic* |
Italic text |
`inline code` |
Inline code |
[link](url) |
Link |
⚙️ Configuration
| Environment Variable | Required | Description |
|---|---|---|
NOTION_TOKEN |
✅ Yes | Your integration secret |
DEFAULT_PRIVACY |
❌ No | private (default) or public |
DEFAULT_PARENT_PAGE_ID |
❌ No | Default parent for new pages |
💡 Examples
Example 1: Bulk Create Database Entries
"Add 20 AI/ML topics to my Topics database with appropriate categories and priorities"
Claude uses bulk_add_rows to create all 20 entries with one command.
Example 2: Daily Workflow
You: "What tasks do I have today?"
Claude: [shows your today_tasks]
You: "Mark 'Review PR #123' as done"
Claude: [calls complete_task]
You: "Add a journal entry: Productive day, finished 5 tasks"
Claude: [appends to journal page]
Example 3: Smart Querying
"Show me high-priority topics that aren't started yet, sorted by category"
Claude builds a complex filter and returns organized results.
🐛 Troubleshooting
"Object not found" errors
The page/database isn't shared with your integration. Fix: Open page → ... → Connections → Connect to your integration.
Empty search results
You haven't shared any pages with the integration yet. Fix: Share at least one page (and its sub-pages will inherit access).
"Cannot create workspace-level page"
Internal integrations can't create pages at workspace root.
Fix: Set DEFAULT_PARENT_PAGE_ID in your config to a parent page.
Rate limit errors
Notion API allows ~3 requests/second. Fix: Server has built-in auto-retry with exponential backoff.
🧑💻 Local Development Setup
If you're contributing or want to run from source, use a .env file instead of the config JSON:
git clone https://github.com/KuldeepJha5176/notion-mcp-server
cd notion-mcp-server
cp .env.example .env
# Edit .env and add your NOTION_TOKEN
uv sync
uv run notion-mcp
Your .env file:
NOTION_TOKEN=secret_your_token_here
DEFAULT_PRIVACY=private
DEFAULT_PARENT_PAGE_ID= # optional
The
.envfile is only for local development. Regular users should use theenvblock in the Claude config JSON as shown in Step 3.
🤝 Contributing
Pull requests welcome! For major changes, please open an issue first.
git clone https://github.com/KuldeepJha5176/notion-mcp-server
cd notion-mcp-server
uv sync
uv run pytest
🔒 Security
- Each user provides their own Notion token
- Tokens stored locally in user's environment
- Server never collects, stores, or transmits user data
- All communication is direct: User → Notion API
- See SECURITY.md for vulnerability reporting
📄 License
MIT — see LICENSE file.
🙏 Acknowledgments
Built with:
- FastMCP — Excellent MCP framework
- notion-client — Notion Python SDK
- Anthropic MCP — The protocol that makes it possible
⭐ Show Your Support
If this server helps you, please:
- ⭐ Star this repo
- 🐦 Tweet about it
- 📝 Blog about how you use it
- 🐛 Report bugs to make it better
Made with ❤️ for the Claude community by Kuldeep Jha
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。