Skip to main content
Technical & Crawl Anthropic Claude

Connect Claude to Google Search Console: MCP and Search Analytics

DEEP ANTHROPIC CLAUDE TECHNICAL WORKFLOW

Human-Edited & Verified • Deterministic AI-assisted research. Editorial standards.

Developer connecting Claude Code to Google Search Console through a local MCP process and OAuth key AI Edited
AI-generated Claude GSC wiring. Raw provenance file.

Last updated: 30 August 2026

Connect Claude to Google Search Console means OAuth to searchanalytics.query (and URL Inspection) from Claude Code or Claude Desktop via MCP or a backend you control—not paste-a-CSV into chat.

TLDR

  • The API is Google’s: POST …/sites/{siteUrl}/searchAnalytics/query (query method).
  • Scopes: webmasters.readonly or webmasters (authorize requests).
  • Claude’s transport is MCP stdio (or HTTP) to a local server that holds tokens. ChatGPT’s transport is GPT Actions (Actions). Same Google API, different client.
  • Indexing API is still JobPosting / BroadcastEvent only (Indexing API).
  • ChatGPT GSC how-to stays on Google Search Console to ChatGPT. Do not copy that H1 here.

Contents

What connecting Claude actually does

Claude cannot see Search Console until a tool call succeeds. MCP exposes tools such as “query search analytics” and “inspect URL.” The model chooses the tool; your server calls Google.

First-party GSC assigned 96 impressions to /claude/seo-google/ and 57 to /chatgpt/seo-google/ in the same 28-day window. Keep distinct H1s. Queries like connect google search console to claude code belong here.

Prerequisites: Search Console permission on the property, a Google Cloud OAuth client or service account, and the Search Console API enabled (prerequisites).

searchanalytics.query from an MCP tool

Engineer inspecting a Search Analytics JSON body with date range and query-page dimensions for Claude AI Edited
AI-generated Search Analytics body. Raw provenance file.

Required body: startDate, endDate. Optional: dimensions, dimensionFilterGroups, rowLimit, startRow. Google does not guarantee every row; results are top rows by clicks.

{
  "startDate": "2026-07-30",
  "endDate": "2026-08-27",
  "dimensions": ["query", "page"],
  "rowLimit": 25000
}

Store refresh tokens on the MCP host. Do not paste client secrets into Claude instructions. Model Context Protocol is the local tool bus (MCP specification).

Claude Code versus ChatGPT Actions

Client How it calls GSC Config surface
Claude Code / Desktop MCP server stdio claude_desktop_config.json / Claude Code MCP
ChatGPT Custom GPT Action OpenAPI + OAuth in GPT builder

Use ChatGPT GSC for Actions. Use this URL for Claude. The skills pillar is the agent-agnostic API notes.

npx @seoskillsai/cli add seo-google --agent claude

Open the Claude hub

FAQ

How do I connect Claude to Google Search Console?

Enable the Search Console API, obtain OAuth or a service account with property access, run an MCP (or HTTPS) wrapper that calls searchanalytics.query, and register that server in Claude.

How do I connect Google Search Console to Claude Code?

Same API. Claude Code adds the MCP server in user or project config. The property URL must match https://example.com/ or sc-domain:example.com.

Is this the Indexing API?

No. Inspection and Search Analytics are webmaster v3. Indexing API publish is limited to JobPosting and BroadcastEvent-in-VideoObject.

Why not one URL for ChatGPT and Claude?

GSC already splits impressions. Shared H1s caused the Perplexity cannibal. Keep transports separate.

Can I paste the service-account JSON into the prompt?

No. The token belongs on the MCP host.

Related reading

Sources

~3,400 Avg. Token Consumption
9s Avg. Execution Latency
$0.009 Estimated API Cost / Run
100% MCP Compliant
RUNTIME CONFIGURATION

Anthropic Claude Configuration Blueprint (claude_desktop_config.json)

Paste this configuration block directly into your claude_desktop_config.json:

claude_desktop_config.json JSON
{
  "google": {
    "autoIndex": true,
    "monitorStrikingDistance": true
  }
}
SEMANTIC ENTITY-ATTRIBUTE MODEL

What the Claude Code Google Search Console & Indexing API Automation Analyzes

Automated diagnostic data points evaluated during every execution run.

Diagnostic Category Specific Data Points Checked Algorithmic Impact
Live GSC Indexation Status Coverage report parsing, crawl anomaly alerts, excluded URLs Detects and resolves indexation blockers within 24 hours
Search Console Query Analytics Impressions, clicks, CTR, position volatility mining Identifies underperforming striking-distance keywords (pos 8–20)
Automated URL Submission Google Indexing API & Bing Submission API batch pings Accelerates new page discovery and indexing by 90%
Sitemap Feed Synchronization Instant sitemap pinging upon new article publishing Ensures immediate Googlebot discovery of fresh content
STEP-BY-STEP WORKFLOW

How to Execute Google Search Console & Indexing API Automation in Anthropic Claude

Imperative configuration instructions with ready-to-run commands.

1

Step 1: Connect Search Console Credentials

Authenticate with GSC service account credentials.

seoskillsai google auth --credentials ./gsc-key.json
2

Step 2: Pull Striking-Distance Keywords

Query performance data to find high-impression, low-CTR terms.

seoskillsai google query --site "https://example.com" --filter "striking-distance"
3

Step 3: Ping Google Indexing API

Submit updated URLs directly to Google for instant re-crawling.

seoskillsai google index --urls "./dist/sitemap.xml"

Make SEO Skills AI a Preferred Source

Keep SEO skill docs highlighted in Google Search and AI Overviews.

Add on Google
Preferred Source added