MCP questions, answered.

The basics

What is the MCP connection? It is an open standard that lets your AI assistant use ScoutRival's tools and read your data from the chat you already work in. The overview walks through what that gives you.

Do I need to write code? No. You paste one config block or click through a sign-in, then ask in plain language. The quick start takes about five minutes.

Which assistants work? Claude (Desktop, web, Code), ChatGPT, Cursor, Grok, the Gemini CLI, and multi-model clients like Cline, Windsurf and Zed. See any client for the full list and how to connect each.

Do the tools do the same as the dashboard? Yes. The read tools return the full composed view you see in the app, not a summary. The tool reference lists every one.

Plans & cost

Which plan do I need? Starter and up. Free plans don't include MCP, and you'll see an upgrade prompt on the token page.

Does using it cost credits? Reading is free. Generating content or running a check costs the same as it does in the app and comes from your shared balance. The credits page lists what each tool charges.

Can I stop it overspending? Yes. Set a daily cap on the token, or keep the token Read-only so it can only read. Both are on the access tokens page.

Safety & control

Is my data safe? Tokens are stored only as a hash, connections are scoped and access-limited, and your competitive set isn't shared with third parties. The security page covers how.

Can it publish to my website on its own? No. Publishing to WordPress always creates a draft you approve, never a live post.

How do I disconnect? Revoke the token at Settings → MCP, or remove the connector in a sign-in client. Both are instant. See troubleshooting if a client keeps a stale connection.

Models & accuracy

Can I use Gemini, DeepSeek or Kimi? Yes. Run the model inside an MCP-capable client and ScoutRival's tools work the same, because MCP is a feature of the client, not the model. The any client page shows which clients carry which models.

Can I trust the AI-visibility numbers? They're measured across two engines — ChatGPT, which is web-grounded, and Gemini, an ungrounded memory probe — and reported with confidence ranges. Read them with their caveats, and treat a position as an average, not a fixed rank. The data page explains how to read them.

Brands, devices & teams

Can one connection cover several brands? Yes, within the token's brand scope. You name the brand when you ask, and the tool works on that brand.

Should each device have its own token? Yes. A token per device lets you revoke one without breaking the others. The access tokens page shows how to mint and manage them.