Skip to main content

What you can ask through MCP

The data your assistant can reach, questions worth starting with, and the limits to know about.

Once connected — by signing in or with an API key — your assistant can reach most of what the dashboard shows. You don't need to know tool names: ask in plain language and it picks the right ones.

What's available

Finding your way around — which brands you track, which prompts, which AI platforms, your tags and competitors.

Visibility and share of voice — how often you're mentioned, how you compare to competitors, your owned-versus-earned split, and a per-prompt brand leaderboard.

Prompt-level gaps — visibility per prompt, filtered by search volume and difficulty. This is the most valuable one, and it's worth knowing it returns two different rates: how often you are mentioned, and how often any brand is. The gap between them separates "competitors are beating me here" from "this question doesn't produce brand answers at all" — completely different problems.

Citations and sources — the domain leaderboard, the breakdown by source and content type, and individual citations with page titles, descriptions and domain authority. Searchable by free text.

The searches the assistants ran — the web queries behind your answers, grouped into topics, readable per AI platform, and the share of answers that involved no search at all.

How you were framed — whether a mention reads as positive or negative, on individual mentions and rolled up per prompt, alongside how much of the range has been scored.

Raw evidence — actual recorded answers, so your assistant can quote what an AI platform really said.

Agent results — ranking advice and video summaries you've already generated.

Questions worth starting with

  • "Which prompts have high search volume, low difficulty, and where am I not being mentioned?" — the shortlist of winnable questions

  • "Compare the sources cited when my brand appears versus when it doesn't." — a diff no dashboard shows directly

  • "Which of my prompts get answered without any web search?" — the ones a page published this week won't reach

  • "Which competitors show up most, and on which prompts do they beat me?"

  • "Summarise my visibility this month versus last, and draft a client update." — where MCP beats the dashboard outright

  • "Pull my ranking advice for the prompts I'm losing and turn it into a content brief."

That last pattern is the real payoff: the data arrives in the same conversation where you do the writing.

Scoping to one brand

If you track several brands, name the one you mean. Otherwise your assistant may ask, or pick the first — and a question spanning brands usually isn't what you wanted.

If you've connected more than one workspace, each is its own connection with its own name in your app. Say which workspace you mean as well.

Limits to know about

  • Date ranges go up to a year.

  • List sizes are still capped per request, but your assistant can ask for the next page from where the last one ended. The response says when there's more, so a first page isn't the whole list.

  • Free-text searches need at least a few characters, and shorter ones are ignored rather than erroring.

  • Adding prompts is capped per request, so a very long list arrives in batches.

Where the dashboard takes over

Two boundaries worth setting expectations on.

It reads agent results; new runs start in the app. Your assistant can pull up ranking advice you've generated, but producing new advice is a click in the dashboard. Agent runs cost credits, and that stays a deliberate decision rather than something an assistant can trigger in passing.

Changes are additive only. With write access — Allow changes ticked when you signed in, or a write-enabled API key — an assistant can add a brand or add prompts. Editing and deleting happen in the dashboard, whatever the connection.

Do note that adding prompts starts regular checking and uses your plan's capacity — no credits, but not free of consequence either. See How limits work.

Did this answer your question?