Search Console MCP

The Search Console connector lets your AI assistant read your own Google Search Console data - clicks, impressions, rankings, index status and sitemaps - and analyse it for you. It is a remote MCP server with built-in sign-in, so in most apps you paste one URL and click Connect. There is no API key to copy. See what it can do for the full tool list.

The server URL

https://toolkitbackend-j69m.onrender.com/api/v1/gsc/mcp

1. Connect Google Search Console

1. Sign in to 500Toolbox with Google (top right). New accounts get free credits.
2. Open Dashboard → Search Console and click Connect Google Search Console.
3. On the Google screen pick the Google account that has access to your sites in Search Console, keep the View Search Console data box ticked and continue.
4. You return to the setup page showing the connected account. Click List my properties to confirm your sites are visible.

2. Add the connector to your AI app

Claude (claude.ai, Claude Desktop and mobile):

1. Settings → Connectors → Add custom connector.
2. Name: Search Console. URL: {{ORIGIN}}/api/v1/gsc/mcp. Click Add.
3. Click Connect. A 500Toolbox page opens - sign in if asked and click Allow.
4. In a chat, open the tools menu and make sure Search Console is switched on.

ChatGPT (plans with developer mode):

1. Settings → Apps & Connectors → Advanced settings → turn on Developer mode.
2. Back in Apps & Connectors click Create. MCP server URL: {{ORIGIN}}/api/v1/gsc/mcp, Authentication: OAuth.
3. Create, then sign in to 500Toolbox and click Allow. In a chat, enable the connector from the + menu.

Claude Code:

claude mcp add --transport http search-console https://toolkitbackend-j69m.onrender.com/api/v1/gsc/mcp
# then inside Claude Code:
/mcp   # choose search-console → Authenticate

Cursor (~/.cursor/mcp.json) - Cursor shows "Needs login"; click it and Allow:

{
  "mcpServers": {
    "search-console": {
      "url": "https://toolkitbackend-j69m.onrender.com/api/v1/gsc/mcp"
    }
  }
}

VS Code with GitHub Copilot (.vscode/mcp.json) - click Start, approve the sign-in, use Copilot Chat in Agent mode:

{
  "servers": {
    "search-console": {
      "type": "http",
      "url": "https://toolkitbackend-j69m.onrender.com/api/v1/gsc/mcp"
    }
  }
}

Apps without OAuth support can send an API key instead. Create one on the API keys page and send it as Authorization: Bearer tb_live_... to the same URL. Stdio-only apps can use the bridge with TOOLBOX_MCP_SERVER=search-console:

{
  "mcpServers": {
    "search-console": {
      "command": "npx",
      "args": ["-y", "@500toolbox/mcp"],
      "env": {
        "TOOLBOX_API_KEY": "tb_live_<your_api_key>",
        "TOOLBOX_MCP_SERVER": "search-console"
      }
    }
  }
}

3. Ask something

• "How did my site do in the last 28 days compared to the month before?"
• "Which keywords am I ranking 5-15 for that I could push onto page one?"
• "My traffic dropped last week - which pages lost clicks and why?"
• "Are my 10 most important pages indexed? Check them."

If you have several properties, name the site in your question ("for example.com") - a bare domain is matched to your property automatically. Apps that show MCP prompts also get five ready-made workflows: Weekly SEO report, Find SEO quick wins, Diagnose a traffic drop, Indexing audit and Content ideas.

What it costs

Troubleshooting

Privacy and disconnecting

Reports are fetched from Google when the AI asks and are not stored. We keep your Google refresh token encrypted (AES-256-GCM) so the connection keeps working, and a log of which tool ran and what it cost. On the setup page you can remove a single AI app, or Disconnect Google entirely - that revokes the token at Google and deletes it. You can also remove 500Toolbox at myaccount.google.com/permissions.