CitedOrNot MCP server

Ask your AI assistant whether ChatGPT, Gemini, Perplexity and other AI engines recommend your brand, which competitors they name instead, and what to publish to get named.

CitedOrNot checks, on a schedule, whether AI engines name your brand when buyers ask what to buy or who to use. This MCP server lets an assistant such as Claude, Cursor or VS Code read those results directly.

This repository is documentation only. The server runs at https://citedornot.com/mcp and is part of the CitedOrNot product; its source code is not published here. This repo exists so you can see what it does and how to connect to it.

What you need

A CitedOrNot account on the Pro or Agency plan (plans and prices).
A personal MCP token: in CitedOrNot go to Settings → MCP and create one. It starts with mcp_.

New to CitedOrNot? The free check needs no signup: citedornot.com.

What you can ask

"Which of my brands is named least often by the AI engines this month?"
"Show the questions where a competitor was named and we weren't, on Perplexity."
"How do the engines differ in tone about ?"
"What should I publish next to get named more often?"
"Which of my agency clients has the lowest visibility?" (Agency plan)

Tools

Tool	What it returns
`list-brands-tool`	Every brand on the account, with question counts and when each was last scanned. Call this first.
`get-brand-visibility-tool`	Visibility rate, recommendation rate, sentiment split, a breakdown by engine and by question, and the competitor gaps.
`list-mentions-tool`	Individual engine answers: the engine, whether the brand was named, tone, how strongly it was recommended, an excerpt and any citations. Filter by question, engine, tone or gaps only.
`get-recommendations-tool`	The current "what to publish" advice: which competitor is winning, the query, proof points and action steps.
`list-clients-tool`	Agency clients with their brands and a rolled-up visibility rate. Appears only for accounts that have clients.

Every tool is read-only. None of them starts a new scan, uses your scan allowance or changes anything on your account. The live list is also published as a server card.

Connect

The endpoint is https://citedornot.com/mcp (streamable HTTP). Send your token as Authorization: Bearer <your-token>, or as X-API-Key: <your-token> if your client can't set an Authorization header. There is no OAuth sign-in.

Claude Code

Terminal

claude mcp add citedornot --transport http https://citedornot.com/mcp --header "Authorization: Bearer <your-token>"

Claude.ai / Claude Desktop

Settings → Connectors → Add custom connector, paste the URL above. When it asks about authentication choose None (this server uses a token, not OAuth), then add a header:

Name: Authorization
Value: Bearer <your-token>

Cursor / VS Code / Windsurf

Add a remote MCP server with:

URL: https://citedornot.com/mcp
Header: Authorization: Bearer <your-token>

Smithery

A listing with a token field is on Smithery: citedornot/ai-visibility.

Limits and access

60 requests per minute.
The token works only while the account is on Pro or Agency. It stops working if the plan is downgraded or you deactivate the token under Settings → MCP.
Treat the token like a password. Anyone who has it can read your brands' results.

Questions

hello@citedornot.com · citedornot.com

Licence

The text in this repository is released under the MIT licence (see LICENSE). That covers this documentation only; the CitedOrNot service and its name and logo are not covered by it.