Use Your AI Visibility Data Inside Your Own AI Assistant: Refora's MCP Server

What is an MCP server?

An MCP server is a standard way to let an AI assistant read data from another service. MCP stands for Model Context Protocol. With Refora's MCP server connected, your assistant can look up your Refora data and answer questions about it in plain language.

Without MCP, using your marketing data in an AI assistant usually means exporting a file, uploading it and hoping the assistant reads it correctly. The data is stale the moment you export it, and you repeat the process every time.

MCP replaces that with a live connection. The assistant asks Refora for the specific data it needs, such as your visibility for the last 30 days or the sources cited for a topic, and uses the answer in its reply. You stay in the conversation; the assistant does the lookup.

Refora's MCP server is available at https://mcp.refora.ai/mcp. The connection is read-only: an assistant can read your Refora data, but it can never change anything in your Refora account. You sign in with your normal Refora email and password, choose which organization to connect, and each connection reads only that organization.

It works with Claude (claude.ai, Claude Desktop and the Claude mobile apps), ChatGPT, Claude Code and other MCP clients that support remote servers. Full setup instructions are in the MCP documentation, and the MCP server product page gives an overview.

What can you ask Refora from your own AI assistant?

You can ask about your AI visibility, share of voice, competitors, citations, sentiment, prompt performance and the AI answers themselves. The assistant picks the right Refora tool, reads the data and explains the result.

The server provides twelve read-only tools, covering:

• Visibility and trends. Your visibility score and how it has changed over time.
• Share of voice. How your mentions compare with competitors.
• Competitors. Head-to-head comparisons and the gaps where competitors appear and you don't.
• Citations. The domains and URLs AI assistants cite, including the most cited pages for a topic.
• Sentiment. How AI assistants describe your brand.
• Prompt performance. How individual tracked questions perform.
• The answers themselves. What each AI model said to your prompts, with the brands each answer mentions and the sources it cites, filterable by prompt, AI model, brand or period.

Examples of questions you can ask:

• "Which websites do ChatGPT and Perplexity cite when they answer questions about our category?"
• "What did ChatGPT actually say when asked about our category, and which sources did it cite?"
• "Show me last month's AI answers that don't mention us."

A few things to know when reading the numbers. Visibility is the share of AI answers that mention your brand, and runs that returned no answer are not counted. Results cover the last 30 days by default, compared with the 30 days before, and you can ask for any period up to 90 days. AI answers come ten at a time, newest first, and long answers are shortened; ask the assistant for the next ones, or to show one answer in full.

How do you connect Refora?

Add Refora as a custom connector in Claude, or as a custom MCP server in ChatGPT, using the server URL https://mcp.refora.ai/mcp. Then sign in to Refora, pick the organization and allow access.

In Claude:

1. Open Customize → Connectors (Settings → Connectors links there) and choose Add → Add custom connector. 2. Name it Refora and enter the MCP server URL. Choose Continue, keep the detected sign-in options and choose Add. 3. Choose Connect, sign in to Refora, pick the organization and choose Allow. 4. In a chat, open the tools menu, make sure Refora is turned on, and ask a question.

On Claude Team and Enterprise, an Owner adds the connector once under Organization settings → Connectors, and each member then connects with their own Refora login.

In ChatGPT:

1. On the web, open Plugins in the sidebar and choose Add → Add custom MCP server. 2. Name it Refora, enter the server URL, keep Authentication on OAuth, tick "I understand and want to continue" and choose Create as a plugin. 3. Choose Connect Refora, sign in, pick the organization and choose Allow. 4. Open Refora under Plugins and choose Try in chat.

On ChatGPT Business and Enterprise, an admin may need to allow the app for the workspace first. You can remove the connector at any time. Each connection can make up to 100 requests per hour. The MCP documentation covers Claude Code, troubleshooting and privacy.

Who is it for?

It's for anyone who already works in an AI assistant and wants their AI visibility data there too: marketers preparing reports, agencies answering client questions and leaders who want a quick answer without opening a dashboard.

You need a Refora account on a paid plan or an active trial. A few typical uses:

• Marketing teams asking "How did our visibility change this month, and why?" while writing an update.
• Content teams asking which sources are cited for a topic before planning new pages.
• Agencies pulling a client's competitor gaps into a conversation while preparing a report, with each connection limited to the organization they chose.
• Leaders getting a plain-language summary without learning a new tool.

The MCP server doesn't replace the Refora app. Tracking, setup and actions still live in Refora. What it adds is a faster way to ask questions about the data you already have, in the assistant you already use.

It also fits the way Refora works overall: track the questions buyers ask, act on what you find, and re-check to see whether it worked. With MCP, the re-check can be a single question to your assistant.

If you don't have a Refora account yet, start with the free AI visibility report to see where you stand today.

Frequently asked questions