How to Connect Claude Code with Notion Using MCP
October 22, 2025
Take Away: At Tag1, we believe in proving AI within our own work before recommending it to clients. This post is part of our AI Applied content series, where team members share real stories of how they're using Artificial Intelligence and the insights and lessons they learn along the way. Building on his earlier post, How AI Helped Me Tame Our Documentation Chaos, Jeff Sheltren, Partner/CIO, shows how integrating Claude Code with Notion’s Model Context Protocol (MCP) enabled deeper analysis and smarter organization of Tag1’s internal documentation, turning a growing content challenge into a connected, intelligent knowledgebase.
Supercharge Your Knowledgebase
Notion is a powerful content repository providing teams the ability to write and share documentation, organize it within powerful database structures, and more. Many teams, like ours at Tag1, use Notion for both internal and external documentation. While Notion's built-in AI can be very useful for things like summarizing meeting notes, it's not designed for larger tasks like analyzing large sets of business data. Leveraging the Notion Model Context Protocol (MCP), you can directly integrate much more powerful LLMs into your team's Notion workspace, unlocking super-powerful AI-based tools that can perform a wide variety of business intelligence, editorial, and analysis tasks.
NOTE: What is MCP?
Model Context Protocol (MCP) is an open standard that allows AI assistants like Claude to securely connect to external data sources and tools. Think of it as a bridge that lets Claude read from and interact with your business applications - whether that's Notion, databases, APIs, or other platforms. Unlike traditional integrations that require custom development, MCP provides a standardized way for AI tools to access your data while maintaining security and control over permissions.
In my case, I was undertaking a rather daunting task: our internal documentation had grown quickly and organically in Notion to the point that it lacked a cohesive navigation structure, causing our employees difficulty in accessing information. Although we had very well written documentation available, it wasn't easy to find it, and it wasn't obvious to users where they should create new content, or how navigation was supposed to flow. You can read the full story on that project in "How AI Helped Me Tame Our Documentation Chaos".
I wanted a way to easily pull large sets of data out of Notion for analysis and potential reorganization, and the Notion MCP provided exactly the tool I needed in order to integrate Claude Code directly with our Notion data to perform that analysis.
Internal Data, Internal Models
At Tag1, we take data privacy seriously. While Notion will allow you to connect to external AI tools like ChatGPT, or Claude, that setup uses external models not under our control. Our policy is that no internal or client code or data can be used on externally hosted LLMs -- for that reason, we host our own internal LLMs. I needed to ensure that Notion data was only sent to those Internal models. Claude Code gives the ability to point to internal LLMs, and so I chose to implement the integration in Claude Code, pointing to our internally hosted models.
Notion MCP Installation and Configuration
Notion provides a hosted MCP server that you can connect to with many AI tools. This guide will focus on Claude Code, but installation with other tools is documented in their MCP documentation. In cases where you need to host the MCP locally, that's possible, but most users will be fine with the Notion-hosted MCP server since your data is presumably already in Notion.
The setup is straightforward: install the MCP within Claude, login via Oauth to your Notion account and grant permissions to a selected workspace.
First, run the following in bash:
$ claude mcp add -s local -t http notion-hosted https://mcp.notion.com/mcpI chose to install it locally, you could also install it directly into a project directory with -s project -- that might be convenient if you're collaborating with others on a shared project.
Once installed, run claude, and it should show the following:
1 MCP server needs auth
Type /mcp to manage MCPsUse the /mcp command as instructed, select the Notion MCP, hit enter and you will open a browser window prompting you to login to Notion if you aren't already, and asking if you want to allow the MCP connection and selecting which workspace to grant it access to.
Once completed, Claude Code will be connected to the Notion MCP and you can use the various MCP tools to view and manage data in your Notion workspace. You can test this in Claude by asking it to retrieve a random page title from your Notion workspace. You can then instruct Claude to interact with Notion with a prompt like:
Find my notes in Notion from the most recent Compliance meeting.That should trigger it to use the MCP to search your Notion workspace. In my case, it found the Notion database containing my Compliance meeting notes, and pulled out a summary for me!
Configure Auto-Allow Permissions in Claude Code
Depending on your needs, you may want to configure Claude Code to auto-allow certain commands through the Notion MCP so that it won't prompt you each time it needs to run a query or fetch a page. If you trust it in your workflow, you may even choose to allow it to write and edit. This can be configured by selecting the "Always Allow" option when Claude goes to run a command. Alternatively, you can configure this by editing .claude/settings.json within your project directory, and add the following contents to allow all search and fetch operations:
{
"permissions": {
"allow": [
"mcp__notion-hosted__search",
"mcp__notion-hosted__fetch"
]
}
}The full list of commands available in the MCP can be viewed by using the /mcp command in Claude Code, selecting the Notion MCP, and then View tools (or just ask Claude). Be careful auto-allowing write and edit until you've tested out your processes -- and even then, be sure there are backups and quality assurance in place. Claude will have whatever level of access you have to contents within Notion, and it would be easy to overwrite or delete things if you aren't paying close attention.
Wrapping Up
With Notion MCP connected, you'll find yourself using Claude for all sorts of tasks you hadn't considered before. Beyond searching for meeting notes, try asking it to analyze patterns across your project databases, generate reports from scattered documentation, or even help reorganize content based on how your team actually uses it.
The setup takes about 10 minutes, but the time savings add up quickly. Instead of clicking through dozens of Notion pages looking for that key information that may be spread out over multiple pages, you can just ask Claude to find it for you. And if you're dealing with larger documentation challenges like we were, this setup becomes the foundation for much more sophisticated analysis.
Curious about how we put AI into action at Tag1? Check out my earlier post, How AI Helped Me Tame Our Documentation Chaos, for a behind-the-scenes look at our approach.
This post is part of Tag1’s AI Applied series, where we share how we're using AI inside our own work before bringing it to clients. Our goal is to be transparent about what works, what doesn’t, and what we are still figuring out, so that together, we can build a more practical, responsible path for AI adoption.
Bring practical, proven AI adoption strategies to your organization, let's start a conversation! We'd love to hear from you.
Image by ken Tomita from pexels