Best Google Search Console MCP Servers in 2026: Hosted and Open-Source Options

Compare Google Search Console MCP servers for Claude, ChatGPT and coding agents, including GSC Magic, GSC Wizard and open-source projects.

TL;DR

The best Google Search Console MCP server depends on how you work. Hosted services such as GSC Magic and GSC Wizard reduce OAuth and maintenance work; open-source servers such as pijusz/mcp-gsc and LukeRenton/google-search-console-mcp give technical teams more control. Compare supported AI clients, actual tools, authentication, reporting depth and the cost of maintaining the connection—not just the number of tools advertised.

What makes a good Search Console MCP server?

A useful MCP server should let an AI assistant ask a real question about a website and return real Google Search Console data. At a minimum, that means property discovery and search performance reporting. More advanced options add URL Inspection, sitemaps, historical reporting or ready-made analysis.

But a long tool list does not automatically make the server better. An integration that returns clear date ranges, handles OAuth safely and works consistently in your chosen AI client is usually more valuable than one with dozens of unreliable endpoints.

The options below are worth evaluating in 2026. Features and plans change, so use each project’s linked documentation before deploying it. This is a practical comparison, not a ranking based on invented speed tests or affiliate commissions.

The shortlist

OptionTypeBest fitMain trade-off
GSC MagicHosted MCP and SEO dashboardFounders and teams wanting Google, Bing and GA4 togetherManaged service and plan limits
GSC WizardHosted SEO and MCP platformTeams wanting a broader multi-source SEO workflowPaid plans and feature-dependent access
pijusz/mcp-gscOpen-source MCP serverDevelopers comfortable with their own Google OAuth setupSelf-maintained runtime and credentials
LukeRenton/google-search-console-mcpOpen-source integrationClaude Code workflows and focused SEO toolsLocal setup and scope depend on the project
muryulia/google-search-console-mcpOpen-source, read-only orientedMinimal reporting with narrow permissionsSmaller, more focused tool surface

The important question is what happens after your assistant can list a property. Can it compare periods, inspect an important page, explain missing queries and show its evidence? Try those tasks before choosing a long-term setup.

GSC Magic: managed MCP with connected SEO sources

GSC Magic is a hosted MCP service with an associated dashboard. Its documented tools cover Search Console analytics, URL Inspection and sitemaps, plus focused reports for quick wins, traffic drops, content decay and cannibalization.

The platform also documents Bing Webmaster Tools, GA4, Core Web Vitals and optional IndexNow actions. That matters if you want an agent to compare Google and Bing visibility or connect search acquisition with website sessions.

Its remote MCP endpoint is https://gscmagic.com/mcp. The documentation explains OAuth and token-based connections for compatible clients, along with read-only tokens and separate write-mode controls.

For a solo founder, the appeal is avoiding a custom OAuth application and maintaining another server. For an agency, the interest may be site limits, group access and reporting. But if all you need is one monthly Search Console CSV, a subscription may not solve a significant problem.

GSC Wizard: another hosted approach

GSC Wizard is an established Search Console-oriented SEO product that also offers MCP connectivity. Its platform and documentation describe broader SEO workflows involving additional sources, including Bing and analytics integrations.

This makes it a reasonable alternative to evaluate when you already want a dedicated SEO workspace rather than a small connector. Compare its currently available tools, AI client compatibility, account limits and included reporting features—not just the name of the product.

GSC Wizard has its own pricing page. Check the plan required for your intended MCP access before budgeting, because trials, packages and payment intervals may differ.

Neither hosted service can reveal privacy-anonymized Search Console queries. Google’s data restrictions still apply, regardless of the product’s AI features.

pijusz/mcp-gsc: open source for hands-on developers

The pijusz/mcp-gsc repository provides an open-source Search Console MCP implementation. It is suitable for someone comfortable running software and configuring Google OAuth credentials.

The attraction is control. You can inspect the code, understand which API calls it makes and operate the integration in your own environment. For a developer who already manages Google Cloud projects, this may be a natural fit.

The trade-off is operational. You are responsible for configuration, credential storage, version updates and keeping the server available. If a client or Google API changes, you may need to investigate that yourself.

Read the repository’s current README and permissions before connecting a production Google account. A GitHub project being public does not, by itself, establish that every dependency or security control has been audited.

LukeRenton/google-search-console-mcp: focused developer workflow

The LukeRenton/google-search-console-mcp project is another option to inspect for Claude Code-oriented use. It combines a Search Console MCP integration with an SEO-focused development workflow.

This approach may suit developers who already work inside a coding agent and primarily want search performance to inform code or content changes.

The key test is whether its tools cover your actual tasks. If you need page-versus-query comparisons and indexed URL checks, confirm both are available in the version you install. If you also need GA4 conversions or Bing data, check whether you will need separate integrations.

As with any self-hosted option, plan for OAuth setup and for maintaining the connection rather than treating installation as a one-time guaranteed event.

muryulia/google-search-console-mcp: smaller read-only footprint

The muryulia/google-search-console-mcp project emphasizes read-only Search Console access. That can be appealing if your agent only needs to inspect properties and performance reports.

A narrower tool surface is not necessarily a disadvantage. Fewer write capabilities can make permissions easier to reason about, particularly in experiments or internal reporting.

The limitation is breadth. If your workflow involves complex report presets, indexing batches, Bing or client-specific dashboards, you may need other tools.

Read the repository’s requirements and supported methods rather than assuming every MCP project exposes the same Google endpoints.

How to choose between hosted and open source

A hosted provider generally handles the remote endpoint, authentication plumbing and service updates. That can save time when connecting from several clients or when team members are not developers.

An open-source implementation may give you deeper control over the server and its data flow. But someone must maintain the runtime, security patches and credentials. It can be inexpensive in cash and expensive in attention.

For a one-person SaaS, ask whether hosting the connector is part of your core product. If it is not, a managed service can be economical. For a company with strict infrastructure rules, self-hosting may be mandatory.

Also examine client support. A local stdio server that works in Claude Code might not be directly usable as a remote connector inside every web application. The transport and authentication model matter as much as the code’s features.

Five tests to run before committing

First, list your actual accessible Search Console properties. The assistant should show full property identifiers, not guess your domain.

Second, request the same seven-day report you can verify in Search Console. Check clicks and impressions for the same property, dates and Search type.

Third, ask for a page-and-query breakdown. Confirm how many rows were returned and whether pagination or provider limits apply.

Fourth, inspect a URL if that is part of your workflow. The agent must distinguish Google’s indexed-version URL Inspection API from a live site test.

Fifth, disconnect or revoke a test credential. A practical offboarding path is part of a reliable integration.

If a provider cannot explain these fundamentals, a glossy list of advanced AI features should not outweigh that concern.

What matters for SEO accuracy

Google Search Console’s Search Analytics API supports clicks, impressions, CTR and position, with dimension choices and row limits documented by Google. The API does not promise every possible query row; rare queries may be hidden.

An MCP server should preserve those limitations. It should not invent missing keywords, treat average ranking as a fixed position or promise that a sitemap submission forces Google indexing.

Demand evidence in every report: property, exact dates, filters, data source and the relevant metrics. Ask the agent to label hypotheses separately from observed changes.

For a complete discussion of the protocol, read our MCP overview.

FAQ

Which Search Console MCP server is easiest to set up?

A hosted server usually requires less infrastructure work than configuring your own Google OAuth client. Actual ease depends on the AI client, account permissions and authentication flow.

Is a free open-source MCP server better?

It may be better if you have the skills and want control. Factor in hosting, updates and credential maintenance when comparing costs.

Can a Search Console MCP server guarantee better rankings?

No. It can provide evidence for decisions, but connecting a tool does not alter Google’s rankings.

Can I connect one server to Claude and ChatGPT?

A compatible hosted remote MCP endpoint can serve different clients when each supports its transport and authentication. Verify each client’s connector requirements separately.

Will all servers return identical data?

Equivalent requests to the same Google property should use the same underlying data. Different defaults, filters, dates, row limits and summary logic can make outputs disagree.

A 15-minute test worth doing before you choose

Use a property you own and write down the expected result from the native Search Console interface. Ask each candidate to list your properties, then request the same seven complete days of clicks and impressions. Confirm the property identifier and Search type match. Next request page-and-query pairs for one URL, and look at the tool’s response rather than only the AI summary.

Finally, revoke the connection and attempt the query again. A managed service should stop returning your private data when its authorization is revoked. Record these results in a simple comparison sheet: setup completed, numbers reproducible, breakdown available, revocation verified, and recurring cost. This is a repeatable evaluation method, not a claim that we benchmarked these services.

For an agency, repeat the property-discovery test with a restricted teammate account. Preventing access to unrelated client sites is more important than saving one or two clicks during onboarding.

A sensible first choice

Choose the smallest reliable setup that answers your actual SEO questions. If you want cross-source reporting without operating OAuth infrastructure, compare GSC Magic and GSC Wizard hands-on. If you want to own the server, inspect the open-source implementations and their maintenance history.

Whatever you choose, keep Google Search Console itself available as your reference. A good MCP server makes its reports easier to use, not harder to verify.

Join the waitlist

Coming soon. Get 30% off Pro for 6 months when you join.

Join the waitlist