CitedOrNot AI Search Visibility

What the CitedOrNot - AI Search Visibility MCP server does

CitedOrNot - AI Search Visibility MCP server lets an MCP-compatible assistant read the results collected by a CitedOrNot account. CitedOrNot checks on a schedule whether selected AI engines mention a brand when users ask buying or recommendation questions. The MCP connection exposes those existing results for analysis; it does not initiate scans or modify account data.

The available data can help answer questions such as which brand has the lowest visibility, where a competitor appears instead, how engines differ in tone, and what content recommendations are currently associated with a brand. Agency accounts can also inspect clients, their brands, and an aggregated visibility rate.

How it works

The service is hosted by CitedOrNot at https://citedornot.com/mcp and uses streamable HTTP. Authenticate each connection with a personal MCP token in either an Authorization: Bearer <token> header or an X-API-Key: <token> header when the client cannot set the Authorization header. OAuth is not used.

Start with list-brands-tool, which returns the brands on the account, their question counts, and the date of the latest scan. Use the returned brand information to inspect visibility or drill into individual mentions. All tools are read-only, so requests consume neither scan allowance nor account-changing operations.

Setup and configuration

An active CitedOrNot Pro or Agency account is required. Create a personal token under CitedOrNot Settings → MCP; the token begins with mcp_. Add the hosted endpoint as a remote MCP server in a compatible client and supply the token as a request header.

Claude Code can be configured with its HTTP transport command. Claude.ai or Claude Desktop uses a custom connector with authentication set to None, followed by an Authorization header containing the bearer token. Cursor, VS Code, and Windsurf use the same endpoint and Authorization header pattern. The token should be treated as a password because it grants access to the account’s brand results.

Tools and capabilities

• list-brands-tool lists account brands, question totals, and the last scan time.
• get-brand-visibility-tool reports visibility and recommendation rates, sentiment distribution, engine and question breakdowns, and competitor gaps.
• list-mentions-tool returns individual engine answers with the engine name, brand-mentioned status, tone, recommendation strength, excerpt, and citations. Filters include question, engine, tone, and gaps-only results.
• get-recommendations-tool provides the current publishing advice, including the winning competitor, query, proof points, and suggested actions.
• list-clients-tool lists agency clients, their brands, and rolled-up visibility. This tool appears only when the account has clients.

Limitations and notes

CitedOrNot - AI Search Visibility is limited to reading results already collected by the CitedOrNot product. It does not run scans, change settings, or update account data. Access is limited to 60 requests per minute, and the token stops working if the account leaves the Pro or Agency plan or the token is deactivated.

The repository documents the service rather than publishing its implementation. The MIT license applies to the repository documentation only; it does not cover the hosted CitedOrNot service, name, or logo. The account’s live scan coverage and results depend on the brands and questions configured in CitedOrNot.