Rankio runs an MCP server at https://www.rankio.studio/api/mcp. Connect it to Claude, ChatGPT, or any MCP-compatible assistant and ask about your AI visibility in plain language — no dashboard, no export, no copy-paste. Setup takes about two minutes: create an API key in Settings, paste one URL into your client. Keys are read-only by default and scoped to your company alone.
The short version: One endpoint, one API key, twelve tools. Read-only by default — reading never spends your quota. Every tool is locked to your own company; no tool accepts a company identifier, so a key cannot be aimed anywhere else. Keys are stored hashed and revoke in under a minute. Included on every plan.
Your visibility data, where you are already working
Most questions about AI visibility are quick ones. Did we move up on that prompt? Who overtook us last week? What should we write next? Answering them usually means opening the dashboard, finding the right view, reading a chart, and carrying the number back to wherever you were actually working.
MCP — the Model Context Protocol — is an open standard that lets an AI assistant read from a data source directly. Once Rankio is connected, your assistant answers those questions from your live data, in the conversation you are already having. If you are drafting a board update in Claude, the visibility numbers come to the draft rather than the other way round.
Nothing is copied or synced. The assistant asks Rankio at the moment you ask the question, and Rankio answers with what is true right then.
What you can ask
These are ordinary questions, not commands. Your assistant works out which tools to call.
“How did our AI visibility score move over the last month, and what drove it?”
“Which of our monitored prompts are we not mentioned in? Who shows up instead?”
“List every competitor that appeared alongside us in the last 30 days, ranked by how often.”
“Pull our content gaps and draft a prioritised brief for the three with the biggest impact.”
“Write the AI visibility section of our monthly client report using this month's Rankio data.”
Connect in three steps
1. Create an API key
In Rankio, open Settings and find MCP Access. Name the key after the device you will use it on, leave permissions on Read only, and click Create API Key.
The key is shown exactly once. Rankio stores only a hash of it, so it cannot be recovered later — copy it before you close the panel. If you lose it, revoke it and make a new one.
2. Add the server to your assistant
Rankio generates the exact command or config block for you when the key is created. For reference:
claude mcp add --transport http rankio \
https://www.rankio.studio/api/mcp \
--header "Authorization: Bearer rk_live_YOUR_KEY"{
"mcpServers": {
"rankio": {
"command": "npx",
"args": [
"-y", "mcp-remote",
"https://www.rankio.studio/api/mcp",
"--header", "Authorization: Bearer rk_live_YOUR_KEY"
]
}
}
}URL: https://www.rankio.studio/api/mcp
Header: Authorization: Bearer rk_live_YOUR_KEY3. Ask something
Restart your client so it picks up the new server, then ask a question about your brand. If the connection is working, the assistant will call a Rankio tool and answer from your data. Start with “What is our current AI visibility score?” — it is the quickest way to confirm everything is wired up.
What the assistant can read
Twelve tools are available. Ten only read; two can trigger work and are hidden entirely unless you deliberately create a key that permits them.
| Tool | What it returns |
|---|---|
get_company_profile | Your brand, website, industry, plan, and remaining quota |
get_visibility_overview | Headline visibility score, confidence level, per-engine breakdown |
list_sov_checks | Recent Share of Voice checks — filter by period, engine, hit or miss |
get_sov_check | One check in full, including the AI's actual answer |
list_prompt_monitors | Tracked prompts with your current position and the competitors present |
get_prompt_monitor_history | Position over time for one prompt — the trend line |
list_url_monitors | Tracked pages and their latest readability score |
get_content_insights | Content gaps, missing topics, and recommendations |
list_citation_reports | Which sources AI engines cite in your category |
list_page_analyses | Past analyses of how an AI agent reads your pages |
The two action tools, available only on a read-and-run key:
| Tool | Effect |
|---|---|
run_sov_check | Runs a new Share of Voice check. Consumes one check from your monthly quota and takes 30–90 seconds |
refresh_visibility_snapshot | Recomputes your score from checks you already ran. No quota cost |
What a key can and cannot do
- Your data only. Every tool is scoped to the company the key belongs to. No tool accepts a company identifier, so there is no argument an assistant could supply to reach another customer's data.
- Read-only by default. A read-only key cannot run analyses and cannot modify anything. The action tools are not merely blocked — they are not listed at all, so an assistant never sees them.
- Reading is free. Reading your data never touches your monthly quota. Only
run_sov_checkdoes, and only on a key you deliberately created with that permission. - Keys are stored hashed. Rankio keeps only a SHA-256 hash. Nobody can read your key back out of the database — not an attacker, not Rankio support.
- Permissions cannot escalate. A key never carries more access than the person who created it. Only admins and auditors can create a key with run permissions.
- Revocation is near-immediate. Revoke from Settings and the key stops working within a minute. Create one key per device so you can cut off a single laptop without disturbing anything else.
- Subscription rules still apply. If a subscription lapses, its keys stop returning data, exactly as the dashboard would.
Common questions
Connect Rankio to your assistant
Create a read-only key in Settings and ask your first question in under two minutes.