ToolsConnector
Notion
Read, search, create and update Notion pages, databases and blocks from Claude or ChatGPT, through Tulina's own MCP server, not Notion's own hosted one.
Official site: www.notion.so
What it can do
1 read/1 write
- 01Read
Read and search pages and databases
- 02Write
Create and update pages and blocks
The connector reads and writes at both levels: individual pages and the databases that structure them. A workflow can update one page, or query a database like a table, depending on which shape holds the source of truth. It works with a Notion internal integration, so it sees only what you share with that integration.
Does Notion have an MCP server?
Yes. Notion hosts its own at https://mcp.notion.com/mcp, with a fallback at /sse for clients that do not speak Streamable HTTP. Sign-in is OAuth in the browser, and only workspace members can connect, not guests. Business and Enterprise workspaces can also issue a token for headless use. Its tools search Notion and connected sources, read and edit pages, query databases with SQL, manage comments, views and files, and reach meeting notes, calendars and Custom Agent sessions. Tulina's connector is the other route, held in the same MCP link as the rest of your tools.
How do you connect Notion to Claude and ChatGPT?
Through Tulina's MCP server, in four steps:
- In Notion, create an internal integration at notion.so/my-integrations and copy its token. For comments, tick the read and insert comment capabilities, which are off by default.
- Share each page or database you want reachable with the integration, from the page menu under connections. An integration sees nothing that is not shared with it.
- Paste the token into Tulina's Notion connector. It is stored in the vault and never shown to the model.
- Add
https://mcp.tulina.ai/mcpto Claude or ChatGPT and sign in with your Tulina account. The MCP page has the same link.
Tulina vs Notion's official MCP server
Notion's server covers the workspace you sign in to. Tulina's covers what you have shared with one integration.
| Notion's MCP server | Tulina | |
|---|---|---|
| Endpoint | mcp.notion.com/mcp, hosted by Notion | mcp.tulina.ai/mcp, one link for every connected tool |
| Sign-in | OAuth as a member, or a token on Business and Enterprise | An internal integration token in Tulina's vault |
| Reach | Everything the signed-in member can open | Only pages and databases shared with the integration |
| Search | Notion pages, databases, users and connected sources | Pages and databases shared with the integration, or every object edited on a given day |
| Reading pages | Fetch by URL or ID | Page properties, blocks, or the whole page as Markdown |
| Editing pages | Create, update, move, duplicate | Create, update, move, append blocks, edit one block, edit as Markdown |
| Databases | SQL queries across data sources, create, update | Query with filters, read the schema, create, update |
| Meeting notes, calendar, Custom Agent sessions | Yes | No tool |
If the model should run SQL across several databases, search connected apps or read meeting notes, Notion's own server is the better choice. Tulina fits when you want the model fenced to a few shared pages, or when Notion is one step in a longer piece of work.
What you can ask it
"What changed in Notion yesterday?" The agent searches with an empty query and the edit day set, and returns every object edited that day, most recent first, in one answer.
"List the rows of the roadmap database where status is To do." The agent reads the database schema, then queries it with a filter and pages through the results with the returned cursor.
"Fix the pricing sentence on the onboarding page." The agent reads the page as Markdown and applies a search-and-replace edit, so the rest of the page is untouched.
Notion pairs with Slack and Gmail in a knowledge base that updates itself, and with Linear in call feedback that becomes tickets and docs.
What the Notion connector can't do
- An empty search can mean nothing is shared with the integration rather than nothing matches. The answer carries a warning that says how to tell.
- A search returns at most 100 objects at a time. The agent follows the cursor for the rest.
- Database tools take the original database or its data source, not a linked view, which is a copy placed in another page.
- Comments need the read and insert capabilities ticked in the integration.
- It has no SQL across databases, no meeting notes, no calendar and no Custom Agent sessions.
- It does not upload files.
- It does not search apps connected to Notion, such as Slack or Google Drive.
Why through Tulina
The connection was never the hard part.
Most systems now publish their own way for a model to reach them, and that is good: it is what Tulina connects to. The hard part is deciding who may use it, what it is allowed to return, and what the model knows before it asks.
The keys stay in one place
YouTeamOrganisation
Connect Notion once. The credential is encrypted in a vault, never held in clear, and never pasted into a chat or a workflow file. A person uses their own key where they have one, their team's where they do not, the organisation's after that, so nobody is blocked and nobody holds more access than they need. Revoke it once and everything running on it stops.
What leaves is filtered before it leaves
fields droppedmasked or hashedpseudonymised
Roles decide who can reach Notion at all. Beyond that, a policy on the connector itself decides what comes back. The data is minimised at the source rather than after it has already reached a model.
Every call is on the record
whatwhowhen
Runs are logged for the organisation. That is what makes an automated action reviewable after the fact, and it is what a security review asks for first.
The run starts knowing your company
namingkey accountsdecisions
A direct connection gives a model access. It does not tell it how your company names things, which accounts matter, or what was decided last quarter. Tulina carries that across every tool, so the same request against Notion returns something usable instead of something generic.
Processes that run on Notion
Call feedback becomes tickets and docs
Every call recorded in the day is distilled overnight, with bugs and feature requests routed to the right team as tickets and docs, plus a Friday digest.
5 steps →Draft the quarterly LP letter
Portfolio metrics gathered from founder updates and dashboards, drafted into your quarterly letter, with every figure carrying the date it was reported.
5 steps →Keep your knowledge base current
An AI knowledge base kept current each morning: yesterday's activity in your tools is read, stale pages are rewritten in place, and a missed day sends an alert.
7 steps →Publish the Monday brief
Product usage, pipeline and feedback themes joined into one weekly executive report, keeping only the metrics that broke trend, posted before the meeting.
5 steps →
