mcpsearchconsole vs Suganthan’s GSC MCP

Suganthan’s GSC MCP has three routes in its Quick start, read on 2 October 2026:

1. Create a Google OAuth client. Every route needs this once for now: create a Google Cloud project, enable the Search Console API, create an OAuth client ID of type Desktop app and download the JSON (3 steps). The README says a built-in Google sign-in with no Cloud steps will ship once Google verifies the shared client.
2. Run the wizard with npx -y suganthan-gsc-mcp setup. It signs you in, checks the connection with a live call, lets you pick a property and writes the config for Claude Desktop and Claude Code. Alternatively, edit the Claude Desktop config by hand, or install the one-click .mcpb bundle from the releases page.
3. Restart the client and ask.

The author's setup guide says "setup takes about fifteen minutes". Indexing API tools need one more step: enable the Web Search Indexing API and have owner-level access in Search Console.

Search Console MCP, the hosted Google Search Console MCP server:

1. Add https://mcpsearchconsole.com/mcp as a connector in your AI client.
2. Sign in with Google and approve the read-only Search Console permission.
3. Ask a question.

That is about a minute, with no Node.js, no Google Cloud project and no config file. The trade-off: requests pass through our server, while Suganthan’s server talks to Google directly from your machine.

What Suganthan’s GSC MCP can do that Search Console MCP cannot

• Stay entirely local. Its README says your data travels between your machine and Google and the developer operates no servers. If no third party may sit in the path, this settles it.
• Deeper ready-made analysis. Tools we do not have include content_decay (three consecutive 30-day periods), traffic_drops (ranking loss versus CTR collapse versus falling demand), content_gaps, topic_cluster_performance, ctr_vs_benchmark, check_alerts and content_recommendations.
• Image search. Eight image SEO tools, from image keyword trends to image_page_audit, which fetches your pages and checks alt text, file names, sizes and image metadata.
• AI conversation queries. genai_conversation_queries classifies AI Mode follow-ups that show up in the query report.
• Submit URLs for indexing. submit_url and submit_batch use Google's Indexing API. We can submit sitemaps (opt-in) but not individual URLs.
• Fresher numbers. It requests dataState: 'all', which includes the most recent, not-yet-final days. Our data trails by about two days.
• Visual dashboards and reports on disk. Results render as cards and charts in Claude Desktop, and generate_report saves a markdown report locally. verify_claim re-checks a number before Claude states it.
• Service accounts and open code. Apache-2.0, so you can audit and extend it.

What Search Console MCP can do that Suganthan’s GSC MCP cannot

• Work outside Claude's desktop apps. claude.ai on the web, ChatGPT, Grok Bot, Cursor, Codex and Mistral Vibe. Its README names Claude Desktop and Claude Code only.
• Need no Google Cloud project. You sign in with Google and that is all, today.
• Site checks. DNS records, redirect chains, robots.txt and TLS certificates (free, unmetered), and PageSpeed with real-user Core Web Vitals.
• Notes and teams. Notes saved in one conversation are there in the next; on Team, they are shared across everyone on your company domain.
• Nothing to maintain. No Node version, npm package or desktop bundle to keep current. Its v2.5.1 changelog fixed token fetches that failed on Node 22 and newer and a desktop bundle that could not start; local servers meet that kind of thing, and hosted ones fix it for you.

Both cover striking distance (it calls them quick wins), cannibalization and low-CTR pages. Neither stores history beyond Google's 16 months, tracks SERP rankings, or reports backlinks or keyword volumes.

Suganthan’s GSC MCP is free; the author puts it as "Total ongoing cost: nothing." What you spend is time: the first setup and the occasional fix after an update to Node, the package or your client. Search Console MCP has a price and very little time cost.

self-hosted yearly cost = (setup hours + upkeep hours per month × 12) × your hourly rate
hosted yearly cost      = monthly price × 12

Example only; replace every number with your own: take the author's estimate of 15 minutes for setup (0.25 hours), assume 10 minutes of upkeep a month (2 hours a year), and call your hourly rate R. That is 2.25 × R a year. Solo costs €59.88 a year before VAT, so in this example they break even at R ≈ €27 an hour. If you already have a Google Cloud project and never hit a broken update, the local server is cheaper; if each person on a team sets it up separately, multiply the time by the number of people.

The full plan details are on our pricing page.

From Suganthan’s GSC MCP to Search Console MCP

1. Add https://mcpsearchconsole.com/mcp as a connector and sign in with the Google account that has your properties.
2. Ask "list my Search Console properties" to check access.
3. Remove the gsc block from your Claude Desktop config (or uninstall the bundle) so Claude does not choose between two similar tool sets.
4. Optional: delete the cached token in ~/.gsc-mcp/ and any Google Cloud project you made only for this.

Plain-language prompts carry over. Prompts that rely on its image, decay or alert tools will need the AI to work from search_analytics rows instead, which takes more calls and gives a less finished answer.

From Search Console MCP to Suganthan’s GSC MCP

1. Create the Google OAuth client and run npx -y suganthan-gsc-mcp setup.
2. Copy any notes you saved with us (ask "list my notes").
3. Remove our connector, cancel any paid plan from your account page and revoke our access in your Google account.

Conversations in claude.ai on the web or ChatGPT will need to move to Claude Desktop or Claude Code. For every option side by side, see best hosted Google Search Console MCP servers.