Connect YouTube to Cursor

Windsor.ai gives Cursor access to 87 YouTube fields. These are some of the groups; the YouTube field reference lists all of them:

• Views and watch time: Video title, Publish date, View Count, Estimated Minutes Watched, Average View Duration, Average View Percentage, Engaged Views.
• Engagement: Likes, Comments, Shares, Subscribers Gained, Subscribers Lost.
• Traffic and audience: Traffic Source, Country, Viewer Age Group, Viewer Gender.
• Video details: Video length, Content Type, Video tags, Playlist Title.

Cursor works best when you ask it to build something with a clear output. Each of these names the fields to use and the file or canvas you want back:

• Pull Video title, Publish date, View Count, Estimated Minutes Watched, Average View Duration and Average View Percentage for videos published in the last 90 days. Build a canvas that ranks videos by minutes watched and marks videos whose average view percentage is below the channel median.

• Pull View Count and Estimated Minutes Watched by Traffic Source for the last 28 days and the 28 days before. Write a Python script that calculates the change per source, and writes traffic_sources.md with the sources that grew or fell most.

• Pull Video title, Subscribers Gained, Subscribers Lost and View Count for this year. Write a script that calculates net subscribers per 1,000 views for each video and saves subscribers.csv sorted by that rate.

• Pull Content Type, Video length, View Count, Average View Duration and Average View Percentage for last month. Build a chart that compares Shorts and long videos, and add a table of the five long videos with the highest average view percentage.

• Pull View Count by Viewer Age Group, Viewer Gender and Country for the last 90 days. Write a monthly audience report as youtube_audience.md and keep the script in the project so I can rerun it each month.

Yes. Cursor can render a dashboard, analysis or report as a canvas next to the chat, and it saves the canvas so you can reopen and rerun it later with fresh data. On paid Cursor plans you can publish a canvas to a read-only share link that members of your Cursor team can open. See the Cursor canvas documentation for the current plan requirements.

No. The YouTube connector is read-only. Cursor reads your YouTube data through Windsor MCP and can write scripts, files and reports in your project, but it cannot create, edit or delete anything in YouTube.

Yes. Windsor.ai supports 350+ data sources, so you can connect YouTube alongside tools such as GA4, TikTok Organic or Instagram and ask Cursor to work with them together. For example, ask Cursor to put YouTube View Count by week next to GA4 Sessions from youtube.com for the same weeks, and build one chart of channel views and site visits.

Open Settings in Cursor Desktop, go to Tools & Integrations → New MCP Server, and add Windsor to mcp.json:

{
  "mcpServers": {
    "windsor": {
      "url": "https://mcp.windsor.ai/"
    }
  }
}

Save the file, click Authenticate next to Windsor in your MCP list, and sign in to Windsor.ai in the browser.

No. The connection takes no code: connect YouTube in Windsor.ai, add Windsor to Cursor’s MCP settings, and sign in. After that you describe what you want in plain language, and Cursor writes and runs the code. You can read and edit everything it produces.

The free forever plan covers 1 data source, so connecting YouTube on its own costs nothing and needs no credit card. Combining YouTube with other sources, such as GA4, needs a paid plan: $19/month for 3 sources or $99/month for 7. Cursor is billed separately by Cursor. See the pricing page.

Yes. These are the resources to start with for the YouTube to Cursor integration:

• YouTube field reference
• Cursor integration documentation
• Windsor.ai connectors for Cursor