Google Search Console MCP Servers Compared: Complete 2026 Guide

Connecting Google Search Console to AI assistants like Claude and ChatGPT has become one of the most practical applications of the Model Context Protocol. But with at least six different GSC MCP servers available on GitHub—plus hosted alternatives—choosing the right one isn't straightforward.

This guide compares every Google Search Console MCP server available in 2026, including setup complexity, features, limitations, and common pitfalls that trip up even experienced developers.

Why Google Search Console + MCP Matters

Before diving into comparisons, let's establish why this integration matters:

Traditional GSC workflow:

1. Log into Search Console

2. Navigate to Performance tab

3. Set date ranges and filters manually

4. Export to spreadsheet

5. Cross-reference with other data sources

6. Repeat for each analysis

With GSC MCP:

1. Ask Claude: "What keywords dropped in position this week?"

2. Get actionable insights in seconds

The difference isn't just speed—it's the ability to ask follow-up questions, combine multiple data points, and get AI-powered analysis without leaving your conversation.

---

Quick Comparison Table

Server	Language	Setup	Auth Method	Tools	Row Limit	Data Freshness	Maintained
ahonn/mcp-server-gsc	TypeScript	Local	Service Account	1	25,000	Final only	✅ Active
AminForou/mcp-gsc	Python	Local	OAuth/Service	19	Default	Final only	✅ Active
Shin-sibainu/gsc-mcp	TypeScript	Local	OAuth	6	25,000	Final only	⚠️ Stale
surendranb/gsc-mcp	Python	Local	Service Account	7	Default	Final only	⚠️ Stale
saurabhsharma2u/search-console-mcp	TypeScript	Local	OAuth	8+	Default	Final only	✅ New
MattiooFR/multi-account	TypeScript	Local	OAuth	6	Default	Final only	✅ New
Ekamoira Open Source	TypeScript	Local	OAuth/Service	6	1,000	Final only	✅ Active
Ekamoira Hosted	TypeScript	Zero	OAuth 2.1	13	1,000	Fresh (all)	✅ Managed

---

Detailed Server Breakdown

1. ahonn/mcp-server-gsc

GitHub: ahonn/mcp-server-gsc
Stars: 157 | Language: TypeScript | License: MIT

Strengths:

• Highest row limit (25,000 rows per request)

• Quick Wins detection with ROI calculations

• Advanced regex filtering support

• Active maintenance with recent Node.js 25 compatibility fix

Limitations:

• Single tool (search_analytics) - no URL inspection or sitemap management

• Service account only - no personal OAuth flow

• Requires manual Google Cloud setup

• Returns "final" data only (2-3 day delay)

Best for: Power users who need maximum data extraction and are comfortable with service account setup.

Setup complexity: Medium-High (Google Cloud project, service account creation, permission grants)

---

2. AminForou/mcp-gsc

GitHub: AminForou/mcp-gsc
Language: Python | License: MIT

Strengths:

• Most comprehensive tool set (19 tools)

• Supports both OAuth and Service Account

• Built-in data visualization (charts, heatmaps)

• Batch URL inspection capability

Limitations:

• Python dependency (requires Python 3.11+)

• Complex setup with virtual environment

• macOS users may face PATH resolution issues

• Default row limits are lower than API maximum

Best for: SEOs who want the most features and don't mind Python setup.

Setup complexity: High (Python environment, OAuth credentials, Claude Desktop configuration)

---

3. Shin-sibainu/google-search-console-mcp-server

GitHub: Shin-sibainu/google-search-console-mcp-server
Language: TypeScript | License: MIT

Strengths:

• URL submission for indexing (unique feature)

• Period comparison built-in

• Automatic retry logic for rate limits

• Japanese documentation available

Limitations:

• OAuth only (no service account option)

• Requires manual refresh token generation

• 2,000 requests/day quota applies

• No mobile-friendly testing

Best for: Users who need URL indexing submission and prefer OAuth flow.

Setup complexity: Medium (OAuth credentials, refresh token generation via browser)

---

4. surendranb/google-search-console-mcp

GitHub: surendranb/google-search-console-mcp
Language: Python | License: MIT

Strengths:

• Clean 7-tool interface

• Dimension/metric discovery tools

• Sitemap management included

• pip installable (pip install google-search-console-mcp)

Limitations:

• Service account only

• Python dependency

• Default API quotas apply

• No URL inspection

Best for: Python developers who want a straightforward, pip-installable solution.

Setup complexity: Medium (Python, service account, environment variables)

---

5. Ekamoira Open Source GSC MCP

GitHub: soumyadeep-ux/gsc-mcp-server
Language: TypeScript | License: MIT

Strengths:

• Both OAuth and Service Account support

• Clean 6-tool interface matching common use cases

• TypeScript (easier debugging than Python for JS developers)

• Actively maintained with hosted alternative available

Limitations:

• Fewer tools than AminForou (no batch inspection)

• Returns "final" data only

• Requires local Node.js setup

• No built-in visualization

Best for: Developers who want full control with TypeScript and the option to migrate to hosted later.

Setup complexity: Medium (Node.js 20+, OAuth credentials, Claude Desktop config)

---

6. Ekamoira Hosted GSC MCP

Endpoint: https://app.ekamoira.com/gsc/mcp
Pricing: 30-day free trial, then included with Ekamoira plans

Strengths:

• Zero local setup - Just enter the URL in Claude.ai/ChatGPT

• OAuth 2.1 with Dynamic Client Registration - No API keys needed for web clients

• Fresh data - Uses dataState: 'all' to match GSC dashboard (not 2-3 day delayed "final" data)

• Server-side date handling - days parameter prevents LLM date hallucination

• URL variant aggregation - aggregationType: 'byProperty' combines www/non-www

• 13 tools - Including mobile-friendly test, comprehensive sitemap management

• Property switching - Select different GSC properties during OAuth consent

• Platform branding - Shows Claude/ChatGPT/Cursor logos in consent flow

• Automatic token refresh - No manual re-authentication

Limitations:

• Requires Ekamoira account (free signup available)

• 30-day trial for GSC-only users

• Hosted dependency (vs self-hosted control)

Best for: Anyone who wants GSC MCP working in 60 seconds without touching Google Cloud Console.

Setup complexity: None (enter URL, authorize, done)

---

New in 2026: Notable Additions

The GSC MCP landscape has grown rapidly since late 2025. AminForou/mcp-gsc has surged to 305 stars (the most-starred GSC MCP server by far), while several notable newcomers have appeared:

saurabhsharma2u/search-console-mcp

GitHub: saurabhsharma2u/search-console-mcp Stars: 6 | Language: TypeScript | Version: 1.9.3

The most actively developed newcomer, shipping rapid iterations since February 2026.

Strengths:

• Site health check tool with week-over-week performance monitoring and traffic anomaly detection

• Built-in Google algorithm updates database (27 entries, 2022-2026)

• Pre-built prompts for Claude, Cursor, and generic MCP clients

• One-command setup: npx search-console-mcp setup

• 100% test coverage on new tools

Limitations:

• Very new (February 2026)---long-term maintenance unknown

• Smaller community vs established options

• No hosted alternative

Best for: Users who want opinionated SEO workflows built into their MCP server.

---

MattiooFR/mcp-gsc-multi-account

GitHub: MattiooFR/mcp-gsc-multi-account Stars: 2 | Language: TypeScript

The first GSC MCP server explicitly designed for managing multiple Google Search Console properties---solving the agency use case.

Strengths:

• Multi-account GSC support in a single MCP server

• Switch between properties without reconfiguring

• Optional Supabase integration for data persistence

Limitations:

• Small community, early stage

• Requires manual OAuth setup per account

Best for: Agencies and consultants managing multiple client GSC properties from a single AI assistant.

---

houtini-ai/better-search-console

GitHub: houtini-ai/better-search-console Language: TypeScript

A unique approach that pulls your entire GSC dataset into a local SQLite database, then exposes pre-built SQL queries as MCP tools.

Strengths:

• Offline analysis---no API rate limits during exploration

• Dashboard generation via Claude Desktop

• Complex cross-referencing queries without API pagination

Limitations:

• Requires initial data sync (can be slow for large properties)

• Data freshness depends on sync frequency

• Early stage

Best for: Power users who want to run complex analytical queries on their GSC data without API limitations.

---

What Changed in the MCP Ecosystem Since 2025

Several structural changes affect how GSC MCP servers work and are deployed:

Governance: Agentic AI Foundation (AAIF) In December 2025, Anthropic donated MCP to the Agentic AI Foundation, a Linux Foundation directed fund co-founded by Anthropic, Block, and OpenAI with support from Google, Microsoft, AWS, and Cloudflare. No single company now controls MCP's direction. For a deep dive into how server discovery works with the new .well-known/mcp.json standard, see our complete guide to MCP server discovery.

Transport: Streamable HTTP Replaces SSE SSE transport is being phased out. The two official transports going forward are STDIO (local servers) and Streamable HTTP (remote servers). If you are deploying a GSC MCP server to the cloud, use Streamable HTTP.

MCP Spec 2025-11-25 Release The one-year anniversary release introduced async tasks (critical for long-running GSC data pulls), improved OAuth via URL-based client registration, and a formalized extensions framework.

Scale MCP now has 97 million+ monthly SDK downloads and over 10,000 active public MCP servers. It is supported by ChatGPT, Claude, Cursor, Gemini, Microsoft Copilot, VS Code, Windsurf, and JetBrains IDEs.

Dormant Servers Note that some original servers have gone dormant: Shin-sibainu/google-search-console-mcp-server (last commit October 2025) and surendranb/google-search-console-mcp (last commit June 2025). If active maintenance matters to your workflow, prefer servers with recent commits.

---

Common Setup Fallacies and Pitfalls

After analyzing GitHub issues, forum posts, and our own implementation experience, here are the most common problems users encounter:

Fallacy 1: "I have Node.js installed, so npx will work"

Reality: macOS applications like Claude Desktop don't inherit your shell's PATH. The infamous spawn npx ENOENT error occurs because Claude can't find npx even though it works in your terminal.

Solutions:

• Create symlinks: ln -s /opt/homebrew/opt/node@22/bin/npx /opt/homebrew/bin/npx

• Use absolute paths in configuration

• Or use a hosted solution that bypasses local setup entirely

Fallacy 2: "Service accounts are easier than OAuth"

Reality: Service accounts require:

1. Creating a Google Cloud project

2. Enabling the Search Console API

3. Creating the service account

4. Downloading JSON credentials

5. Adding the service account email to each GSC property manually

6. Setting environment variables correctly

For personal use, OAuth is often simpler despite the refresh token dance.

Fallacy 3: "The API returns the same data as the GSC dashboard"

Reality: By default, the Search Console API returns "final" data only—which is 2-3 days behind the dashboard. To get fresh data matching what you see in GSC, you need to specify dataState: 'all' in your API requests.

Most open-source servers don't do this. The Ekamoira hosted server does.

Fallacy 4: "I can query any date range I want"

Reality: Google Search Console data has hard limitations:

• 16 months maximum - Data older than 16 months is deleted

• 2-3 day freshness lag - Unless you request dataState: 'all'

• Row limits - 1,000 in UI exports, 25,000 maximum via API per request

Fallacy 5: "LLMs handle dates correctly"

Reality: LLMs are trained on historical data and may default to dates from their training period (e.g., 2024). When you ask for "last month's data," the LLM might literally request November 2024 instead of the actual last month.

Solution: Use relative date parameters like days=28 that calculate dates server-side. The Ekamoira hosted server specifically implements this to prevent date hallucination.

---

Feature Comparison Deep Dive

Authentication Methods

Server	OAuth 2.0	Service Account	API Key	Zero-Setup
ahonn/mcp-server-gsc	❌	✅	❌	❌
AminForou/mcp-gsc	✅	✅	❌	❌
Shin-sibainu/gsc-mcp	✅	❌	❌	❌
surendranb/gsc-mcp	❌	✅	❌	❌
Ekamoira Open Source	✅	✅	❌	❌
Ekamoira Hosted	✅ (OAuth 2.1)	N/A	✅	✅

Tool Coverage

Capability	ahonn	AminForou	Shin-sibainu	surendranb	Ekamoira OS	Ekamoira Hosted
Search Analytics	✅	✅	✅	✅	✅	✅
URL Inspection	❌	✅	✅	❌	✅	✅
Sitemap List	❌	✅	✅	✅	✅	✅
Sitemap Submit	❌	✅	❌	✅	✅	✅
URL Submission	❌	❌	✅	❌	❌	❌
Mobile Test	❌	❌	❌	❌	❌	✅
Quick Wins	✅	❌	❌	❌	❌	❌
Batch Inspect	❌	✅	❌	❌	❌	❌

---

Which Server Should You Choose?

Choose ahonn/mcp-server-gsc if:

• You need maximum data extraction (25,000 rows)

• You're comfortable with service account setup

• Quick Wins detection is valuable for your workflow

• You don't need URL inspection or sitemap management

Choose AminForou/mcp-gsc if:

• You want the most comprehensive feature set

• You're comfortable with Python environments

• You need batch URL inspection

• Data visualization is important

Choose Shin-sibainu/gsc-mcp if:

• You need URL indexing submission

• You prefer OAuth over service accounts

• You want automatic retry logic

• Period comparison is a frequent task

Choose Ekamoira Open Source if:

• You want TypeScript for easier debugging

• You need both OAuth and service account options

• You want the option to migrate to hosted later

• Full self-hosted control is important

Choose Ekamoira Hosted if:

• You want to be running in 60 seconds

• You don't want to manage OAuth credentials

• Fresh data (not 2-3 day delayed) matters

• LLM date accuracy is important

• You'll query from Claude.ai, ChatGPT, or Cursor

---

Getting Started

Option 1: Self-Hosted (Any Open Source Server)

1. Clone the repository

2. Set up Google Cloud project with Search Console API

3. Create OAuth or service account credentials

4. Configure environment variables

5. Add to Claude Desktop config

6. Troubleshoot PATH issues if on macOS

Option 2: Hosted (Ekamoira)

1. Visit app.ekamoira.com/tools/gsc

2. Sign up and connect your Google Search Console

3. Add https://app.ekamoira.com/gsc/mcp to Claude.ai or ChatGPT

4. Authorize when prompted

5. Start querying

---

Conclusion

The Google Search Console MCP ecosystem has exploded in 2026, growing from six implementations to over 20 active projects on GitHub. For developers who want maximum control and don't mind setup complexity, open-source options like ahonn/mcp-server-gsc and AminForou/mcp-gsc offer powerful capabilities.

For everyone else—especially those who've fought with spawn npx ENOENT errors or OAuth credential management—hosted solutions eliminate the friction entirely.

The best server is the one you'll actually use. If setup complexity means you never get around to connecting GSC to your AI workflow, a zero-setup hosted option delivers more value than a feature-rich server sitting in a GitHub star list.

---

Sources

• ahonn/mcp-server-gsc - TypeScript GSC MCP with Quick Wins

• AminForou/mcp-gsc - Python GSC MCP with 19 tools

• Shin-sibainu/google-search-console-mcp-server - TypeScript with URL submission

• surendranb/google-search-console-mcp - Python pip-installable

• soumyadeep-ux/gsc-mcp-server - Ekamoira open source

• Google Search Console API Limitations - Data retention and quotas

• MCP Setup Troubleshooting - Common PATH issues