Google Search Console

Connect Google Search Console to bring first-party organic search data into a CrawlConsole property. The CrawlConsole MCP server can then query performance, identify opportunities, inspect URLs, and read submitted sitemap status without exposing Google credentials to the agent.

Connect a Search Console site

1. Open the CrawlConsole property you want to configure.
2. Open Settings, then Data Connectors.
3. Select Connect Google Search Console and approve the requested Google access.
4. Choose the Search Console domain property or URL-prefix property that represents the CrawlConsole property.
5. Confirm that Search Console appears as connected.

You need permission to edit data sources in the CrawlConsole workspace and access to the selected site in Google Search Console. A Search Console connection belongs to the CrawlConsole property and its workspace. Agent keys use their own scopes and optional domain restrictions to determine which workspace properties they can query; access is not inherited from the individual Google account that originally completed OAuth.

Give an agent access

Create an Agent API key under Settings and API Keys. Grant properties:read so the agent can call list_properties, and grant telemetry:read for the Search Console tools.

Always begin with list_properties and pass its canonical propertyId to the other tools. The GSC tools do not accept a Search Console site URL, OAuth connection id, or access token. CrawlConsole resolves the connector from the authorized workspace and property on every call.

Call list_properties, find example.com, and use its propertyId to return
Google Search Console opportunities for the last 28 complete days.

Available MCP tools

get_gsc_performance

Query clicks, impressions, CTR, and average position. Group results by any supported combination of date, hour, query, page, country, device, and search appearance. You can also apply AND filters, select a Google search surface, include fresh data, compare with the immediately preceding period, and paginate with nextCursor.

{
  "propertyId": "PROPERTY_ID_FROM_LIST_PROPERTIES",
  "days": 28,
  "dimensions": ["query", "page"],
  "searchType": "web",
  "comparePreviousPeriod": true,
  "limit": 100
}

The tool defaults to the last 28 complete UTC days and 100 rows. A response can contain up to 1,000 rows; pass the returned opaque cursor back with the same query to continue. Explicit startDate and endDate values use inclusive YYYY-MM-DD dates.

get_gsc_opportunities

Return a prioritized layer derived from current and previous Search Console performance. It can identify:

• striking-distance queries within a selected position range;
• high-impression results with low CTR;
• query and page pairs whose clicks declined, including pairs that disappeared from the current period;
• possible cannibalization where multiple pages receive impressions for the same query.

The thresholds are inputs, so an agent can tune them to the site instead of relying on hidden scoring rules. The tool defaults to a 28-day comparison and returns up to 200 opportunities.

{
  "propertyId": "PROPERTY_ID_FROM_LIST_PROPERTIES",
  "days": 28,
  "types": ["striking_distance", "declining", "cannibalization"],
  "minImpressions": 100,
  "minPosition": 4,
  "maxPosition": 20,
  "limit": 50
}

inspect_gsc_url

Inspect one absolute HTTP or HTTPS URL inside the connected Search Console property. The response includes available index verdict, coverage, robots, fetch, canonical, last-crawl, mobile-usability, and rich-result fields.

{
  "propertyId": "PROPERTY_ID_FROM_LIST_PROPERTIES",
  "url": "https://example.com/docs/getting-started"
}

For a URL-prefix Search Console site, the URL must share its origin and path prefix. For a domain property, the root domain and its subdomains are allowed. CrawlConsole rejects out-of-property URLs before asking Google to inspect them.

get_gsc_sitemaps

List submitted sitemap paths, last submission and download times, pending state, warnings, errors, and submitted/indexed counts by content type. This tool is read-only: it does not submit or delete sitemaps.

{
  "propertyId": "PROPERTY_ID_FROM_LIST_PROPERTIES",
  "includeContents": true,
  "limit": 100
}

Access, usage, and freshness

Search Console calls are owned-property queries. They do not consume CrawlConsole's metered backlink or external-intelligence Agent allowance, including when that allowance is exhausted. A generous per-key concurrency safety limit still applies, and Google may temporarily limit a busy Search Console property. If a call is limited, retry with fewer parallel requests.

CrawlConsole briefly caches identical GSC reads to reduce latency and unnecessary Google API calls. Search Console performance itself can be delayed or incomplete; use dataState: "all" when fresh data matters, and check the returned incomplete-date metadata before treating the newest rows as final.

Search Analytics returns Google's top rows for a query rather than a guaranteed exhaustive export. Use focused dimensions and filters, and paginate when you need deeper result sets.

Troubleshooting

• Property not found: call list_properties again and use the canonical id available to the current Agent key.
• Search Console is not connected: connect it from that property's Data Connectors settings.
• Connection does not match: reconnect and select the Search Console site for the same domain as the CrawlConsole property.
• Authorization unavailable: reconnect Google Search Console; CrawlConsole may no longer have permission to refresh access.
• Inspection URL is outside the property: use a URL covered by the selected domain or URL-prefix property.
• Temporary limit: reduce parallel calls and retry shortly.

The page is part of the dedicated documentation sitemap. Publishing a CMS override or updating this Markdown source refreshes the page, navigation, agent exports, search index, and sitemap entry together.