I Built an MCP That Turns Keyword Clusters Into SEO Content Calendars

I have run the same loop hundreds of times.

Keyword Insights finishes a clustering job. You get 5,000 keywords, 2,600 clusters, 1,700 topic hubs, intent and funnel data on every row. Then you lose an afternoon turning any of it into something your content team can actually ship on Monday.

3 hours later the usual result. 12 calendar items in a spreadsheet, 4 colours of highlighting, and a nagging feeling that something important is buried in row 1,847.

So I built another MCP. It takes the clustering CSV and outputs a scheduled content calendar in 3 minutes.

What it does

2 tools, each with a specific job.

parse_clustering_csv reads the CSV, validates 18 expected columns, compacts the top 500 clusters by opportunity, enriches each one with derived columns (ranking keyword counts, intent mismatch counts, hub sizes, supporting keywords), and hands the whole thing to your AI with a detailed decision framework baked in.

export_content_calendar takes the JSON calendar your AI generates and writes a formatted Excel workbook with 5 sheets, plus a flat CSV. Ready to drop into ClickUp or hand to your content team.

No API keys to manage. Claude (or whichever AI you use) runs the strategy. The MCP handles the parsing and the export.

The 2 things a spreadsheet cannot do

The full writeup on the Keyword Insights blog covers 5 or 6 use cases. Two of them are the reason I built this.

Intent mismatches. The cluster ranks at position 14 because Google wants a how-to guide and you are ranking a product page. A spreadsheet tells you “position 14.” It will not tell you why 14 is the ceiling. This MCP aggregates the is_intent_match field across every ranking keyword in the cluster. If more than 60% come back mismatched, the cluster gets flagged as CREATE with the reason written out in plain English.

Striking distance decisions. Every SEO knows to look at clusters ranking between position 16 and 30. The question nobody answers automatically is whether the existing page should be refined or rewritten from scratch. The MCP uses the cluster_content_score field (a 0 to 1 measure of how well a page covers its cluster’s topic) to make the call. A score above 0.4 means UPDATE. Anything lower is a REWRITE. The “polish vs start over” decision sits in the output instead of in your head.

Where it slots in the stack

My MCP chain now runs like this. GSC MCP pulls the query level organic data. BigQuery MCP takes over when API sampling limits start to matter. Prophet MCP handles forecasting. This one turns a Keyword Insights clustering CSV into a schedulable plan.

Each one does a specific job. None of them pretend to be a full SEO platform. Claude sits in the middle and picks which tool to call based on the question you ask.

Get it and the deep dive

The repo lives at github.com/Suganthan-Mohanadasan/kwi-content-calendar-mcp/. Clone, npm install, npm run build, point your Claude Desktop config at the built binary. 2 minutes total.

The detailed walkthrough is on the KWI blog. I wrote it there because that is where most of the people who would actually use this tool already spend their time. It covers the decision framework the AI uses, the prioritisation maths, a worked example on a real 5,060-row dataset, and the full Excel output breakdown.

Read the full writeup on the Keyword Insights blog.

If you do not already use Keyword Insights, you will not have a clustering CSV to feed this. The MCP is built around the specific schema KWI exports, which is not an accident. KWI does SERP-level clustering at scale, which is the hard part. This MCP just makes the output useful on a Monday morning.