The Model Context Protocol is the open standard for letting an AI assistant read live data from a tool. Connect llemmy once, then ask about your AI visibility from inside Claude or Cursor. No dashboard tab, no export, no SQL.
"Ask your AI assistant how your brand's visibility is trending." Your assistant calls the right llemmy tools and answers, with no query language to learn.
The server can only read. It's scoped to your API key, so an assistant sees exactly your data and nothing else, and can never change a thing.
A flagship developer and agency differentiator: llemmy is the GEO platform you can talk to from your own AI assistant.
In llemmy, open API Keys and generate one. It scopes the MCP server to your data.
Point Claude, Cursor or any MCP client at the llemmy server and paste your key. One config block, shown opposite.
"How is my brand's visibility trending?" "Which sources does AI cite most?" Your assistant does the rest.
{
"mcpServers": {
"llemmy": {
"url": "https://llemmy.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Your AI client picks the right tool for the question and stitches the results into an answer. Every one is read-only.
Lists every project in your workspace, the starting point for any question about a specific brand.
Returns the prompts llemmy runs against the engines for a project, the questions you're tracking your brand against.
Pulls the actual answers each engine gave, so your assistant can read, quote and summarize what ChatGPT, Claude, Gemini and the rest said.
Returns visibility and mention metrics for a brand across the engines, the numbers behind "how am I trending?"
Surfaces the pages AI engines cite most when they answer about you, the sources to win, audit or reclaim.
Giving an AI assistant access to a tool should never mean giving it the keys. llemmy's MCP server can only read. It cannot run prompts, change settings, delete data or touch billing. And because it authenticates with your API key, it is scoped to your account and the projects you can access, and never reaches another tenant's workspace.
Pull GEO data into Cursor or Claude without leaving your editor. Ask for the numbers as you build the report, dashboard or content brief.
Ask your assistant to summarize visibility across every client project at once, then turn the answer into a client-ready update in seconds.
Interrogate the AI responses themselves: quote the exact passages, find the cited sources, and explore patterns conversationally.
Add teammates, plus free read-only viewer seats so every client can watch their own brand.
One visibility score across every engine, the data the MCP server reads.
Positive and negative, with the exact passages and the sources the AI cited.
Everything llemmy tracks across the AI answer layer, in one place.
Sign up, generate an API key and connect llemmy to Claude or Cursor in minutes. The AI models are included, no extra keys to wrangle.