Rubric MCP server

Rubric by GoGoChimp is a hosted MCP (Model Context Protocol) server. Connect it to an AI assistant that supports remote MCP servers and the assistant can read your Rubric audits and act on them: open a scorecard, pull the ranked fixes, start a crawl, check a draft before you publish, get the exact change for one failing check, and re-crawl the pages you changed to see whether the check now passes.

What it is

Rubric estimates how citable each page of a site is for AI search engines: ChatGPT, Perplexity, Google AI Overviews, Gemini, Copilot and Claude. The MCP server gives your own assistant access to those results. Rubric's engine computes every score. The MCP only returns what the engine produced and queues new work for it, so your assistant reads the numbers and never makes one up. The scores in your assistant match the ones in your Rubric dashboard.

Endpoint and transport

Get a free API key

Authentication

Every tools/call request needs the key in the Authorization header as Bearer <key>. initialize, tools/list and the prompts work without a key, so a client can show the tool list before you add one. A tool call with a missing, unknown or revoked key returns JSON-RPC error -32001.

The server doesn't support OAuth yet. Clients that can't send a custom header, and only offer OAuth or no authentication for remote servers, can't connect to it for now. That includes custom connectors added under Settings, then Connectors, in claude.ai and the Claude apps, and connectors in ChatGPT.

Connect your client

In each snippet below, replace YOUR_RUBRIC_API_KEY with your key. rubric is the local name your client shows for the server, and you can change it.

Claude Code

Run this in a terminal:

claude mcp add --transport http rubric https://rubric.gogochimp.com/api/mcp --header "Authorization: Bearer YOUR_RUBRIC_API_KEY"

Check it with claude mcp list. The prompts appear as /mcp__rubric__fix and /mcp__rubric__help.

Claude Desktop

Claude Desktop's config file starts local servers, so connect through the open-source mcp-remote bridge. It needs Node.js. Open Settings, then Developer, then Edit Config, and add:

{
  "mcpServers": {
    "rubric": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://rubric.gogochimp.com/api/mcp",
        "--header",
        "Authorization:${RUBRIC_AUTH}"
      ],
      "env": {
        "RUBRIC_AUTH": "Bearer YOUR_RUBRIC_API_KEY"
      }
    }
  }
}

The key sits in env rather than args, because Claude Desktop on Windows can mangle spaces in args. Restart Claude Desktop afterwards. The config file is claude_desktop_config.json, in ~/Library/Application Support/Claude/ on macOS and %APPDATA%\Claude\ on Windows.

Cursor

Add this to ~/.cursor/mcp.json to use it in every project, or to .cursor/mcp.json for one project:

{
  "mcpServers": {
    "rubric": {
      "url": "https://rubric.gogochimp.com/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_RUBRIC_API_KEY"
      }
    }
  }
}

VS Code

Add this to .vscode/mcp.json in your workspace. VS Code asks for the key the first time the server starts and stores it for you, so the key never sits in the file:

{
  "inputs": [
    {
      "type": "promptString",
      "id": "rubric-api-key",
      "description": "Rubric API key",
      "password": true
    }
  ],
  "servers": {
    "rubric": {
      "type": "http",
      "url": "https://rubric.gogochimp.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ${input:rubric-api-key}"
      }
    }
  }
}

ChatGPT

ChatGPT's custom connectors use OAuth or no authentication. They don't send a custom Authorization header. Rubric needs that header, so ChatGPT can't connect until Rubric supports OAuth.

Other clients

Any client that can reach a remote Streamable HTTP server and send a custom header works. Give it the endpoint above and the header Authorization: Bearer <key>. If a client can only start local (stdio) servers, use the mcp-remote bridge as in the Claude Desktop example.

Tools

Generated from the server's own tool definitions. Read tools only look at your existing audits. Tools that queue a job run on Rubric's engine and show up on your account, and some of them fetch live pages. None can delete or change anything you already have.

Tools that queue a job answer straight away with a job_id. Poll crawl_status until the status is done, then read the audit or the result. check_draft checks the text or HTML you pass it and doesn't fetch URLs. To check a live page, use start_crawl. verify_fix re-crawls up to 10 pages per run, and explain_answer compares up to 6 cited pages with yours.

Limits by plan

Using the MCP is free on both plans. What differs is how much work it can start:

A day is a rolling 24 hours. If start_crawl is not given a page count it asks for 500, capped at your plan's limit. See pricing for everything else that differs between the plans.

Privacy and what is sent

• To Rubric: the arguments of each tool call. That means the URLs you ask it to crawl, check or compare, the questions you pass to explain_answer, and the draft text you pass to check_draft, which is stored with that job on your account. Rubric also records basic usage events against your account (for example, that a crawl was started over the MCP) and the time your key was last used.
• What Rubric fetches: crawls fetch public web pages, as a normal web client, only for the tools marked "fetches pages" above.
• Your key: stored only as a one-way hash, scoped to your account alone, and revocable at any time on the MCP page.
• What comes back: only your own audits and jobs. A key can never read anyone else's data.
• Your AI client: the results go to the assistant you connected, and that provider's own terms cover what you send through it. Rubric's own model never scores your site and never trains on your data.

The full policy is in the privacy policy. See also the terms.

What the score is, and what it isn't

The score is an estimate built from the evidence on your pages: content, structure and technical set-up. It doesn't count the citations you already have, and it doesn't promise any. Rubric shows where your pages stand and what to change first. The engines still make the decision. How the score works explains the method.

Support

Email info@gogochimp.com. For a broader introduction, read connecting the MCP in the help centre.