Search data for the
assistant you already use

A read-only MCP server for Claude, ChatGPT, Codex and Claude Code. Six sources, thirty-seven tools, and an explicit answer when the data runs out instead of a confident guess.

  • Read-only, enforced in code
  • No API keys to paste
  • Free, limits published

Sources

Six sources, not two.

Most connectors stop at Search Console. That leaves you unable to answer the two questions that matter most: did those clicks become anything, and what is the competition actually ranking for?

Google Search Console

Every query, page and country you already own.

Clicks, impressions, CTR and position by query, page, country, device and search appearance. Index status, canonicals, crawl details, sitemaps.

7 tools · Search Console API + URL Inspection APIDetails

Bing Webmaster Tools

The second index, with its crawl health.

Daily traffic, query and page stats, crawl health with HTTP status groups, sitemaps and per-URL index detail.

6 tools · Webmaster Tools API accessDetails

Analytics 4

Clicks, next to what those clicks earned.

Sessions, key events and revenue alongside search clicks — including a bridge tool that puts both datasets side by side on the same page.

2 tools · Analytics Data APIDetails

PageSpeed + CrUX

Lab scores and real field data, kept apart.

Lighthouse lab audits per URL, and Chrome UX Report field data for real Core Web Vitals. Missing field data is reported as missing, never as good.

2 tools · None — public endpointsDetails

Competitor engine (ours)

What the competition ranks for, without a subscription.

Live Google result pages for any keyword, plus Bing’s own monthly search counts. Builds keyword gaps, share of voice and rival movement without buying a keyword database.

4 tools · Free Programmable Search engineDetails

Monitors and sharing

Six checks that run whether you ask or not.

Six scheduled checks with in-app, email and webhook delivery. Every evaluation is stored, so you can diff two runs and share a snapshot as a public link.

6 tools · NoneDetails

Comparison

Same category, different ceiling.

ScopeMCP is measured against the best-known commercial connector in this space. It is a fair comparison on sources and pricing, and the differences below are all things we actually ship.

CapabilityScopeMCPTypical paid connectorNotes
Google Search ConsoleyesyesPerformance, sitemaps, URL inspection.
Bing Webmaster ToolsyesyesCrawl health and keyword research.
Analytics 4yessoonIncluding a clicks-to-conversions bridge.
PageSpeed + CrUX field datayesyesLab and field, kept clearly separate.
Competitor keyword gapyesnoBuilt in-house, no keyword database purchase.
URL diff between two datesyesnoRanking loss versus click-through loss.
Scheduled monitors and alertsyesnoEmail and webhook, with stored history.
Score with visible weightsyesnoUnmeasured components excluded, not zeroed.
CSV / JSON export and share linksyesnoRevocable public links per snapshot.
PricefreepaidNo card, no plan, published rate limits.

The comparison reflects public documentation as of writing. Features change; if something here is out of date, we would rather hear about it than leave it standing.

Connect

One endpoint. Four agents.

Everything is served from ${env.appUrl}/mcp over Streamable HTTP, with discovery metadata at /.well-known/mcp.json so a client can find it from the domain alone.

remote MCP endpoint
https://scopemcp.galavic.world/mcp
Claude Code
{
  "mcpServers": {
    "scopemcp": {
      "type": "http",
      "url": "https://scopemcp.galavic.world/mcp"
    }
  }
}

Limits, published

The numbers before you hit them.

Most tools hide their ceilings until you trip them. These are in the server instructions too, so your agent knows them without having to ask.

  • Rate limits protect ScopeMCP and the upstream providers. They are not a usage meter: there is no plan and no billing.
  • The competitor engine is limited by the Google Programmable Search free tier of 100 queries per UTC day. Cache hits and previously fetched result pages do not consume it.
  • Cached reports do not count against the free SERP quota, and neither do scheduled monitors that reuse a stored snapshot.
  • Every report response states which window was searched, which calls were made and whether the answer came from cache.
Search data tools60 / 60s
URL inspection20 / 60s
SEO analysis12 / 60s
Competitor engine10 / 60s
Free SERP quota100 / day

All limits are per workspace and shared across every agent and credential on it. Cached reports do not consume upstream quota.

Connect it, then ask the question you have been avoiding.

Three minutes to connect. No card. Bring your own OAuth app if you would rather not use ours, or run the whole thing yourself — the schema, the tools and the read-only guarantee are all in the repository.

try this first
Show me every Search Console
property I can access, then give me
the SEO health score for the last 28
days and explain each component.