Notion Meeting Intelligence

Prepare meetings with Notion context, tailored templates, and Codex research.

Notion Meeting Intelligence is an official OpenAI skill from the curated skills repository. According to the provider, it helps Codex prepare meetings with relevant Notion context, tailor agendas and pre-reads to attendees, and add focused research to the preparation. The skill is not a standalone transcription application and it does not replace Notion. It is an operational guide for an agent that searches Notion through the intended MCP tools, fetches important pages, and turns that source material into a reviewable meeting foundation.

Bringing Notion context together

The workflow starts by clarifying the meeting objective, attendees, date, time, duration, and decisions that should be possible by the end. The agent then searches Notion for previous notes, project specifications, OKRs, open action items, and existing decisions. Key pages are fetched individually afterward. This makes the preparation traceable instead of producing an agenda from generic model knowledge. Blockers, risks, and open questions are surfaced early, so they can be resolved before the meeting rather than discovered in the room.

Selecting a meeting structure

The skill maps different objectives to suitable templates. A status update needs a different structure from a decision or approval meeting, sprint planning, a retrospective, a one-on-one, or a brainstorming session. Templates in the official repository provide reusable sections for context, desired outcomes, timeboxes, owners, discussion questions, decision criteria, risks, and follow-up actions. The agent is expected to adapt a template to the actual situation instead of copying it mechanically.

Building an agenda and pre-read

Using the gathered context, the agent drafts an agenda or pre-read and can create a new Notion page for it. Source links are included, topics receive owners and expected outputs, and available time is distributed across the agenda. For decision meetings, the structure separates options, impact, risks, and recommendation. For planning meetings, dependencies and preparation tasks can be made explicit. The result is therefore more than a polished invitation: it is a working document that supports collaboration before, during, and after the meeting.

Research and maintenance

When outside context is useful, Codex can add concise research about industry conditions, benchmarks, risks, or best practices. According to the provider, claims should carry source links and facts should remain distinct from opinion. If the goal, attendees, or constraints change, the agent updates the Notion page and can maintain a brief changelog when there are multiple revisions. Follow-up tasks created by the meeting can be linked to the relevant Notion database.

Requirements, safety, and boundaries

Practical use requires Codex to have a connected Notion MCP integration. The official skill describes adding the Notion endpoint, enabling the remote MCP client, and completing OAuth login. After a successful login, Codex may need to be restarted. If the connection is unavailable, the agent should pause rather than inventing documents or decisions. Because Notion pages may be read and changed, permissions, target pages, and proposed edits need review before saving. Users should take particular care with confidential notes, personal information, and internal decisions, ensuring that they remain inside the intended workspace. Preparation quality also depends on the available Notion content, source freshness, and human oversight of the resulting recommendations.

Positioning and source quality

The skill is published in the official OpenAI repository as MIT-licensed source material. OpenAI’s Notion page additionally describes searching and referencing Notion pages and creating or updating databases for roadmaps, risks, and action items. Those statements are attributed here as provider information. The skill’s GitHub page remains the primary source for its concrete workflow. The repository contains instructions, templates, and examples, but it cannot guarantee that every organization has identical Notion permissions, page structures, or meeting conventions.

Installation instructions

This guide follows the official OpenAI skill and describes a controlled workflow from connection setup to a finished meeting page.

Step 1: Prepare the connection

First confirm that Codex is available in the environment and that enabling remote MCP is acceptable for the workspace. Add the official Notion MCP endpoint through the Codex configuration, then complete OAuth login in the browser. Do not place credentials in chats or project files. Codex may need to be restarted after login before the Notion tools become available.

Define the goal and source scope

Begin every preparation with the meeting objective, attendees, duration, date, and intended decision. Ask Codex to search for previous notes, project pages, OKRs, specifications, and action items. Review the fetched pages yourself and keep sensitive material within the approved scope. A sound preparation names its sources instead of presenting an apparently certain summary without evidence.

Step 2: Choose a template and create the page

Select a template that matches the purpose, such as status, decision, planning, retrospective, one-on-one, or brainstorming. Have Codex fill in context, goals, timeboxes, owners, risks, open questions, and expected outputs. Before creating a Notion page, check the title, target workspace, and permissions before saving.

Review research

Ask Codex for additional research only when it materially helps the decision or preparation. Each external claim should have a verifiable source link. Keep facts, assumptions, and recommendations in separate sections. Do not accept a recommendation automatically when the Notion sources are stale or incomplete.

Maintain changes and tasks

When the meeting changes, update the existing page deliberately and verify the result in Notion. Follow-up tasks can be linked to the relevant database. A short changelog helps when there are multiple revisions. Finish with human review of links, owners, deadlines, confidential content, and the decision the meeting is actually meant to produce.

Free
Provider
OpenAI
License
MIT
Last reviewed
09.09.2026

Repository and documentation

Categories

Compatible with

Codex