MCP tool reference

Server and sign-in

Setting up a key and your tool is on Connect an AI tool. A project id, finding id or scan id from another account answers as not found.

Tools

list_sites

List projects. Every project (monitored site) on the account with its health, schedule, latest scan, open counts and connections. Start here to find project ids.

It takes no inputs.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "list_sites",
    "arguments": {}
  }
}

get_site_overview

Project overview. Plain-English status, open fix and grow counts by priority, latest scan and how fresh each evidence source is, for one project.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_site_overview",
    "arguments": {
      "site_id": "[SITE_ID]"
    }
  }
}

list_findings

List findings. Website and provider findings for one project, grouped by shared check (a shared check is not a proven shared cause). Use get_finding with a finding_id for evidence and guidance.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "list_findings",
    "arguments": {
      "site_id": "[SITE_ID]"
    }
  }
}

get_finding

Finding detail. Evidence, guidance (what to do, why it matters, how it clears), a copyable developer brief and a derived timeline for one finding and the group that shares its check.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_finding",
    "arguments": {
      "site_id": "[SITE_ID]",
      "finding_id": "[FINDING_ID]"
    }
  }
}

list_opportunities

List growth opportunities. Search Console growth ideas for one project: query, URL, clicks, impressions, position and estimated upside. Metrics are from the data Lookout received, not complete site traffic; estimates are scenarios, not forecasts.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "list_opportunities",
    "arguments": {
      "site_id": "[SITE_ID]"
    }
  }
}

get_search_totals

Search totals. Current and previous 28-day clicks, impressions, CTR and position plus eight weekly click totals from the latest Search Console sync or import.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_search_totals",
    "arguments": {
      "site_id": "[SITE_ID]"
    }
  }
}

list_scans

List scans. Recent website scans for one project, newest first. Scans that did not finish never mark anything as passed. explanation says why a scan stopped or skipped pages; reasons lists each reason with its URL count and up to 10 example URLs. Scans recorded before reasons have neither; an unfinished one keeps only detail.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "list_scans",
    "arguments": {
      "site_id": "[SITE_ID]"
    }
  }
}

get_scan_digest

Scan digest. What changed in one complete scan compared with the previous complete scan: new, reopened (came back), still open (persistent), passed, excluded and unobserved (not checked again: still open, never passed) findings, plus page changes, with examples capped per category: up to 2,000 for new, reopened, passed, addedPages and unobservedPages, and 20 for the rest. Digests with version 2 also carry health and coverage; checks_passing is a percentage (0–100, rounded down to one decimal) of passing / (passing + failing + unverified) over website check keys, with excluded checks and advisory suggestions left out, and is null for older digests and for a scan that counted no checks. It is not a search ranking or traffic measure. capped, when present, counts pages per check that failed beyond the limit of 1,000 findings per check, so no finding records them; page evidence keeps each URL’s latest scan only, so it shows their results until a later scan records those pages again. cappedLabels names each of those checks as page evidence does. explanation and reasons, when present, say why the scan stopped or skipped pages, as list_scans does.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_scan_digest",
    "arguments": {
      "scan_id": "[SCAN_ID]"
    }
  }
}

get_url_inspection

Stored URL Inspection. The latest stored Google URL Inspection report for up to 10 URLs. Reads Lookout’s stored reports only; it never calls Google or requests indexing.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_url_inspection",
    "arguments": {
      "site_id": "[SITE_ID]",
      "urls": "[URLS]"
    }
  }
}

get_core_web_vitals

Field Core Web Vitals. Stored Chrome UX Report (real-user) p75 LCP, INP and CLS by URL or origin scope and device, with collection window. Separate from lab measurements.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_core_web_vitals",
    "arguments": {
      "site_id": "[SITE_ID]"
    }
  }
}

get_lab_performance

Lab performance. The latest stored PageSpeed lab run and up to 10 recent attempts for one URL, per device. Reads stored results only; it never runs PageSpeed.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_lab_performance",
    "arguments": {
      "site_id": "[SITE_ID]",
      "url": "[URL]"
    }
  }
}