Skip to main content

What MCP is and why you'd use it

Connect MentionBird to Claude, ChatGPT or another AI app and ask questions about your own visibility data.

MCP lets you connect MentionBird to the AI assistant you already use, so you can ask questions about your own data in plain language instead of clicking through a dashboard.

Ask Claude "which prompts is Nike losing where Hoka is winning, and what's cited in those answers?" and it queries your account and answers. No exporting, no spreadsheets.

What it's actually good for

A dashboard is good at showing you a number. It's worse at answering a question that spans several screens — which is where this earns its place.

Cross-cutting questions. "Which of my prompts have high search volume, low difficulty, and low visibility?" touches three things at once. One question, one answer.

Comparisons the UI doesn't lay out. "What sources get cited when Nike is mentioned, versus when it isn't?" That diff is genuinely useful and no dashboard view shows it directly.

Turning data into work. Once your assistant has the findings in context, the next request is natural: draft the brief, write the outreach email, summarise it for a client. That's the real reason to connect it.

Recurring reporting. Ask the same question every month and let your assistant write the summary.

How it works

You add one address to your AI app and sign in with your MentionBird login when it asks. You choose which workspace to connect, click Allow, and your assistant can then call a set of MentionBird tools — visibility, competitors, citations, searches, prompts, agent results — and read the answers.

There's no key to copy and no configuration file to edit. Your app stays signed in and renews its access in the background.

One connection reads one workspace. The sign-in screen asks which, and your assistant only ever sees that workspace's data. To work across several, connect once per workspace.

Every app connected this way is listed under Connected apps in MentionBird, where you can disconnect it at any time — see Seeing and disconnecting connected apps.

Cursor, Custom GPTs, Gemini and your own scripts connect with an API key instead. Same tools, same data — the key simply takes the place of signing in.

What it can change

Almost everything is read-only. A connection can read your data but change nothing unless you allow it: tick Allow changes when you sign in, or create an API key with write access. Even then, changes are limited to two things: adding a tracked brand, and adding prompts. Both only ever add.

Editing and deleting happen in the dashboard, never through MCP. That's a deliberate design choice — a connected app, or a leaked key, has no way to remove anything from your workspace.

It also reads agent results rather than starting new runs. Your assistant can pull up ranking advice you've already generated, but starting a new one is a click in the dashboard. Agent runs cost credits, and spending them stays a decision you make deliberately.

One nuance worth knowing: adding prompts through MCP does start regular visibility checking, which uses your plan's capacity. It costs no credits, but it isn't inert.

Who it's for

You don't need to write code. If you can paste an address into your AI app's settings and sign in, you can set this up.

It's most useful if you're already working with an AI assistant daily, or if you're managing several brands and want to ask questions across them without clicking through each.

Did this answer your question?