AI agents are already doing operator work: checking dashboards, drafting schema, summarizing competitors. The missing piece has been a live connection to AI visibility data — whether ChatGPT, Claude, Gemini, and Perplexity actually recommend your business.
Scope ships a remote MCP server for that. This guide covers what it is, how to connect it, what your agent can do, and how it sits next to your Google SEO program rather than instead of it.
What is Scope MCP?
Model Context Protocol (MCP) is the open standard that lets assistants like Claude, ChatGPT, and Cursor call external tools. Scope runs a Streamable HTTP MCP server at:
https://scope.online/mcp
Once connected, the agent can use 15 tools: read your visibility score, list citations, compare competitors, run a readiness check, and draft fixes. Reads are free. Actions that spend checks tell you the cost first.
This is the same product surface as the Scope app — not a toy demo. Agent access is included with Growth.
Why this matters for SEO teams
Google SEO is not going away. Keyword research, technical crawls, and content still drive qualified traffic. Scope does not replace that work.
What has changed is the second answer surface: people ask ChatGPT or Claude who to hire. Those answers are not your Google ranking report. An agent connected to Scope can:
- Pull this week's AI visibility score the same way it already pulls Search Console
- Flag citation gaps that sit beside — not instead of — your backlink profile
- Draft FAQ and schema that help both Google rich results and LLM answers
Keep SEMrush / Ahrefs / your SEO stack for Google. Use Scope MCP for AI recommendations. That is the complementary setup.
How to connect
Full install tabs live on Scope for AI agents. Short version:
Claude
Settings → Connectors → add Scope, or paste https://scope.online/mcp as a custom connector with a restricted API key (read + scan:run).
Try: “How visible is my business in AI search?”
ChatGPT
Settings → Connectors → add Scope → enable it in the composer tools menu → sign in.
Try: “Pull my Scope visibility report and summarize what changed this week.”
Cursor
Add to .cursor/mcp.json:
{
"mcpServers": {
"scope": {
"url": "https://scope.online/mcp",
"headers": {
"Authorization": "Bearer YOUR_SCOPE_API_KEY"
}
}
}
}
Create keys at app.scope.online/settings/agents.
What the agent can do
See the tool reference for schemas. In plain language:
See (always free): plan and remaining checks, businesses, visibility report, prompt results, citations, competitors, recommendations, readiness score.
Do (says cost first): add a business, run tracked prompts, test one prompt, start/stop monitoring.
Fix (never writes to your site): draft schema/FAQ, mark a recommendation done.
Nothing publishes to your website without you.
Example workflows
- Monday review — “Pull my Scope visibility report, compare it to last week, and summarize what changed.”
- Readiness sprint — “Run a readiness check, pick the highest-impact fix, and draft it.”
- Citation audit — “Which sources does ChatGPT cite when it mentions us — and which are we missing?”
- SEO pairing — “Given this Search Console landing page and my Scope citation gaps, what should we publish next that helps both Google and AI answers?”
That last prompt is the point: the agent should treat SEO and AI visibility as one operating system.
How this compares
Profound and Peec have started publishing MCP / agent workflows. Scope’s difference is a production MCP endpoint plus a free AI visibility checker so you can see the gap before you buy. Details: Scope vs Profound and Scope vs Peec.
Safety and permissions
Default key scopes are read and scan:run. Optional: business:write. Revoke any key from settings. See agent security.
Get started
- Run the free AI visibility checker.
- On Growth, open agent settings and create a restricted key.
- Connect Claude, ChatGPT, or Cursor using the install tabs.
- Ask: “How visible is my business in AI search — and what is the top fix?”
Your SEO program keeps running. Your agent now has the AI-recommendation layer too.