Google Results & AI Overview MCP server

How an agent would call it

Two steps: search returns refs, then data_fetch opens one of them. The arguments below are the arguments that were sent.

{
  "tool": "search",
  "arguments": {
    "query": "best crm for small business",
    "sources": [
      "google-serp"
    ],
    "limit": 3,
    "runId": "<from run_begin>"
  }
}

{
  "tool": "data_fetch",
  "arguments": {
    "sourceId": "google-serp",
    "ref": {
      "id": "ai-overview:best crm for small business"
    },
    "runId": "<from run_begin>"
  }
}

This runs against an account — send Authorization: Bearer <token> with the call. Get one at app.bedrocktap.com/dashboard/tokens; the free tier needs no card, and it reaches google-serp like every other source.

What it costs. 10.5 credits per query — one call, one page, whatever the page contains. The search above returned an AI Overview and 3 organic results for 10.5 credits; a fetch costs nothing.

What came back

{
  "ref": {
    "id": "ai-overview:best crm for small business",
    "title": "Google AI Overview"
  },
  "text": "The best CRM for a small business depends on your specific needs, with Zoho CRM standing out as the top overall choice for value and customization.Zoho CRM: Best for budget-conscious teams and custom features.…",
  "fetchedAt": "2026-09-12T15:04:30.286Z"
}

One page came back as 4 refs: the AI Overview first, then three organic results in position order — pcmag.com at 1, zendesk.hk at 2, slack.com at 3. The AI Overview ref carried content and sources, the pages Google credited, each with its own title and description; one of them was a Reddit thread. Each organic ref carried position, displayedUrl and description.

Observed 2026-09-12. One call against a live source — ask again tomorrow and the rows move.

What a row contains

The first row is the AI Overview, when Google shows one: no URL, and meta carrying the answer text and the sources Google credited. It is surfaced as its own row rather than buried in metadata, because for many queries it is what Google now shows instead of the links. Every row after it is one organic result, carrying its rank, the URL as displayed and the snippet. One call returns one page; ask again for page two. A fetch returns what the page said about a result, never the body of somebody else’s site.

A question it answers. Who ranks for the query my buyers type, and what does the AI Overview tell them instead?

Routing guidance

What an agent reads to decide whether google-serp is the right source, and what to reach for when it is not.

Answers

Three questions for the price of one: who ranks organically, what the AI Overview says and which sources it cites, and what related questions people ask. Competitor ranking checks, AI Overview citation tracking, and keyword discovery.

Query shape

A plain search query.

Price

Priced in credits, per row, per query or per call depending on the source. Every call is quoted before it runs — `cost_estimate` returns the exact figure and spends nothing. See /pricing.

Latency

Fast.

Prefer instead

Use this over `ai-visibility` when the question is about Google specifically: roughly five times cheaper and forty times faster, and Google’s AI Overview is where most AI search volume actually sits.

Cannot answer

Search volume or keyword difficulty (use `google-keywords`), ChatGPT and Perplexity citations (use `ai-visibility`), or backlinks.

Caveat

An AI Overview is not always present — informational queries usually trigger one, commercial queries often do not. Absence is a real finding, not a failure: no AI Overview result is emitted when none was shown.

Related sources