Search Console MCP: GSC data for your agent · Rankdelta

A Search Console MCP lets your AI agent read your Google Search Console data — clicks, impressions, CTR, position, top queries and pages. RankDelta's also pairs it with rankings and AI visibility, so the agent sees Google and AI in one connection.

Quick answer

A Search Console MCP is a Model Context Protocol server that lets an AI agent read your Google Search Console data — clicks, impressions, CTR, average position, top queries and top pages — so you can ask questions in plain language instead of clicking through GSC or exporting CSVs. RankDelta ships one (disclosed: it's our product) as part of its hosted, open-source MCP at mcp.rankdelta.ai. What makes it worth using over a GSC-only connector is that the same agent connection also sees your rankings and your AI visibility — so instead of "here are your top queries," you get "here are the queries you already earn clicks on, where you rank for them, and whether ChatGPT names you for the same topics." One connection, the full picture: Google and AI.

What the agent can read

RankDelta's MCP exposes two Search Console tools, both read-only:

get_gsc_property — the connected GSC property for a site (reads the project cache; no OAuth tokens are exposed to the agent).
get_gsc_analytics — a quick Search Analytics overview for a site: clicks, impressions, CTR, average position, top queries and top pages, over a 7 / 28 / 90-day window (RankDelta's synced copy, instant).
query_gsc_search_analytics — live Search Analytics over any range in the last 16 months (all Google keeps), by query, page, country, device, date or search appearance, with filters and up to 25,000 rows: a query's clicks month by month, the pages that lost impressions, Discover or image traffic.
inspect_gsc_url — URL Inspection for one page: indexed or not and why, last crawl, robots.txt, Google's canonical vs yours, sitemaps, mobile usability and rich results.
list_gsc_sitemaps — submitted sitemaps with their status, errors and warnings.

Two honest notes so you know exactly what you're getting: every call is read-only with your own Search Console grant, and GSC's "average position" is Google's own metric, not a third-party rank tracker — for tracked-keyword positions you'd use RankDelta's rank tools alongside it. Manual actions, security issues and the full Page indexing report are not in Google's API, so no MCP can read them; check those in Search Console.

The prompt you'd actually paste

Connect RankDelta. Pull my Google Search Console top queries and top pages for the last 28 days. For the top 10 queries, tell me which ones I already get clicks on but rank below position 5, and — using AI visibility — whether ChatGPT or AI Overviews name me for those same topics. Where I get impressions but no clicks, flag the page to improve.

What comes back

The agent calls get_gsc_analytics (plus list_ranks and get_ai_visibility for the cross-reference) and returns something like:

Query	Clicks	Impr.	CTR	Avg. pos	You in AI answers?
ai seo software	41	1,900	2.2%	6.3	competitor named
ai visibility tool	28	1,100	2.5%	4.1	you named
generative engine optimization	12	2,400	0.5%	11.8	nobody

…plus your top pages by clicks and the "impressions but no clicks" list — which is the title/description-and-intent to-do. The last column is the part a plain GSC MCP can't give you: it's from RankDelta's AI-visibility layer, so the same view shows where Google sends clicks and where AI sends recommendations.

Why pair Search Console with AI visibility

GSC tells you what Google already clicks. It cannot tell you whether an AI assistant recommends you — and a growing share of buyers now ask ChatGPT or read an AI Overview before they ever reach a blue link. Reading GSC in isolation optimizes for half the funnel. RankDelta's MCP puts both in one connection:

GSC → the queries and pages Google actually sends traffic to.
AI visibility → whether ChatGPT, Perplexity, Gemini and AI Overviews name you for those same topics (AI Share of Voice).
Rankings + keywords → where you sit and what to target next (SEO MCP).

That cross-reference — "I earn Google clicks here, but AI recommends a competitor" — is the brief a GSC-only tool can't write.

When NOT to use it

You want raw, live GSC API access. Ours is a cached, agent-friendly read. If you need real-time Search Analytics API pulls in your own code, use the GSC API directly.
You only ever need GSC and nothing else. If rankings and AI visibility aren't part of your workflow, a single-purpose GSC connector (or the GSC UI) is simpler.
You haven't connected a property. get_gsc_analytics returns data only for a site with GSC linked in RankDelta first — connect it in the app, then the agent can read it.

FAQ

What is a Search Console MCP?

A Search Console MCP is a Model Context Protocol server that exposes your Google Search Console data — clicks, impressions, CTR, average position, top queries and pages — so an AI agent can read it directly. RankDelta’s does that and pairs it with rankings and AI visibility in one hosted, open-source connection.

Does it read live Search Console data?

It reads RankDelta’s cached copy of your Search Analytics (synced from the connected property), not a fresh live API call on every request. That keeps it fast and keeps OAuth tokens away from the agent. Windows are 7, 28 or 90 days.

Is GSC position the same as a rank tracker?

No. GSC’s "average position" is Google’s own aggregate metric for your property. For tracked-keyword SERP positions, use RankDelta’s rank tools (list_ranks, check_ranks) alongside GSC.

Do you also expose Google Analytics?

Yes — the same MCP exposes GA4 property and analytics tools, so an agent can read your GA4 data over the same connection. GSC and GA4 together give the agent both search and behavior data.

Which agents can connect?

Any MCP-compatible client — Claude (desktop and Cursor), ChatGPT connectors, Cursor, Grok. One OAuth connection; no API keys to paste.