DEV Community

Cover image for How I Automated Keyword Opportunity Finding in Cursor using GSC + MCP (Free Workflow)
glq0546
glq0546

Posted on

How I Automated Keyword Opportunity Finding in Cursor using GSC + MCP (Free Workflow)

#mcp #seo #ai #cursor

If you're doing SEO in 2026 and you're still exporting CSVs from Google Search Console (GSC) just to paste them into an AI... we need to talk.

It's slow, you hit the 1,000-row limit, and you have to manually guess which keywords are worth targeting.

I wanted a way to query my GSC data directly inside Cursor using natural language. So I built a workflow combining Cursor Rules (for the logic) and the Model Context Protocol (MCP) (for the data).

Here is how you can set it up in 5 minutes and find your "striking-distance" keywords instantly.

Step 1: The SEO Expert Cursor Rule
First, we need Cursor to act like a Senior SEO Expert. Create a .cursorrules file in your project root and paste this in:

text

Role: Senior SEO Growth Engineer

You are my SEO consultant. All judgments must be based on real GSC data.
Never hallucinate numbers. If there is no data, ask for it.

Constraints

  • Query real data using the gsc_opportunity_radar tool.
  • Filter out brand terms and navigational keywords automatically.
  • Focus on keywords ranking between position 5 and 20 with high impressions.

Scoring Formula

score = estimatedClicks × intentWeight × positionMultiplier
(transactional = 1.2, commercial = 1.15, informational = 1.0)

Output Format

  1. Opportunity Table (sorted by score, max 10 rows)
  2. Top 3 Actionable Prescriptions (Title rewrite, Page edits, Internal links)
  3. Excluded Keywords List (to verify brand filtering) Step 2: Connect GSC to Cursor via MCP Cursor can't natively read GSC data. We need a bridge. I built a free, remote MCP server for this so you don't have to deal with complex local OAuth setups.

Go to gsc-mcp-core.vercel.app and log in with Google.

Click "Generate API Key" and copy the JSON config.

Add it to your ~/.cursor/mcp.json file:

json
{
"mcpServers": {
"gsc": {
"url": "https://gsc-mcp-core.vercel.app/api/mcp",
"type": "streamableHttp",
"headers": {
"Authorization": "Bearer gsc_live_YOUR_API_KEY"
}
}
}
}
Restart Cursor, and you should see the gsc-mcp server turn green in the settings.

Step 3: The "Aha!" Moment
Open a new chat in Cursor and ask:

"Scan mysite.com for striking-distance keywords from the last 28 days. Filter brand terms and give me the top 3 prescriptions."

Cursor will automatically call the MCP server, pull your live GSC data, and generate a beautiful Markdown table with a step-by-step action plan. No more Excel, no more CSV tab-switching.

Why this is a game-changer:

Zero Setup: No Python scripts, no local Docker.

Contextual: The AI knows your CTR baselines (Top 3 = 28%, Rank 4-6 = 12%, etc.) and filters the noise.

Free Beta: You can scan a whole site for a week within the free limit (50 requests/day).

If you want to test it out, grab an API key here: https://gsc-mcp-core.vercel.app

Let me know in the comments what SEO workflows you automate with MCP! 👇

Top comments (0)