CiteRank

The AI search visibility platform that tells you why AI won't cite your brand, then hands you the fix.

View on citerankscore.com

Documentation

Citerank MCP Server

Citerank exposes its audit suite as a Model Context Protocol server. Connect from Claude Code, Claude Desktop, Cursor, or any MCP-capable agent and run AI visibility audits, generate schema, check brand citations, and simulate agent journeys, without leaving your editor.

Connect your client

Most clients sign you in to Citerank directly: add the server URL, then sign in and click Allow when your client asks. No key to copy. Every tool call runs as your account, on your plan.

https://citerankscore.com/api/mcp-server

Claude app (claude.ai, Claude Desktop, Claude mobile)

1

Open Settings → Connectors → Add custom connector.

2

Name: Citerank. URL: https://citerankscore.com/api/mcp-server. Leave the advanced OAuth fields empty.

Claude Code

claude mcp add --transport http citerank https://citerankscore.com/api/mcp-server

Then run /mcp in Claude Code and choose Authenticate for citerank.

Cursor, VS Code and other clients

Add this to the client's MCP configuration (Cursor: Settings → MCP → Add new MCP server, or .cursor/mcp.json). The client offers the sign-in the first time it connects:

{
  "mcpServers": {
    "citerank": {
      "type": "http",
      "url": "https://citerankscore.com/api/mcp-server"
    }
  }
}

Using an API key instead

For clients without MCP sign-in, scripts and CI, create a key in Settings → Developer API (it starts with crk_live_ and is shown once). Send it as the header Authorization: Bearer YOUR_KEY, or, for clients that only take a URL, put it on the end: https://citerankscore.com/api/mcp-server?key=YOUR_KEY. Treat that URL like a password.

Managing access

Signing in creates an API key named for the client, for example "Claude (connected 2026-11-12 14:03 UTC, 9f2c)". It is listed in Settings → Developer API with your other keys; revoke it there to disconnect that client.

Check it works

Ask your assistant: "Use citerank to check the agentic readiness of example.com." If the client says sign-in is required, finish the sign-in it offers, or check the key if you connected with one.

Discovery

Clients that discover servers automatically find Citerank through its AI Catalog at https://citerankscore.com/.well-known/ai-catalog.json, which lists the Server Card at https://citerankscore.com/api/mcp-server/server-card (MCP Server Cards, SEP-2127). The card gives the endpoint and the protocol versions it supports. Sign-in is discovered the standard way: an unauthenticated request gets a 401 pointing to https://citerankscore.com/.well-known/oauth-protected-resource/api/mcp-server. It is also listed in the MCP Registry as io.github.WebforMike/citerank.

Available Tools

citerank_analyze_url

Run a full AI visibility audit on a URL. Returns schema analysis, AI citation score, E-E-A-T signals, agentic readiness, and prioritized issues.

url (required) · focus: full | schema | citations | agentic | eeat

citerank_check_brand_citations

Check whether a brand or domain is cited in AI-generated answers (Google AI Overviews, Gemini). Returns citation rate, share of voice, top cited queries, and competitor analysis.

domain (required) · keywords: optional array, max 10

citerank_generate_schema

Generate JSON-LD structured data for a URL. Analyzes page content and produces schema markup ready to implement.

url (required) · schemaType: optional, auto-detected if omitted

citerank_check_agentic_readiness

Check how ready a website is for AI agent interactions. Tests MCP endpoint, WebMCP declarative tools, potentialAction schema, A2A agent card, llms.txt, and more.

url (required)

citerank_simulate_agent_journey

Simulate an AI agent attempting to complete a task on a website. Returns step-by-step results and specific fixes for each failure point.

citerank_ai_traffic

AI traffic measured at your own server: visits from ChatGPT, Perplexity, Claude, Gemini and other assistants, pages AI reads, top landing pages and orders from AI visitors.

domain (required) · days · from / to · compare: previous | year | none · platform · page

citerank_list_projects

Your projects, and those shared with you, with each one's latest Citerank Score, grade and date, competitors and your role.

none

citerank_get_project

One project in detail: its Citerank Score history and the latest run of every tool on it, with scores and dates, as in your Audit History.

domain (required)

citerank_get_site_audit

The latest Site Crawler audit for a project: pages crawled, site health, AI readiness, schema coverage and the most severe issues, each with its page. Reads the stored audit; it does not start a crawl.

domain (required) · limit (default 15, max 50)

citerank_list_wp_pages

List the pages and posts on your connected WordPress site, with whether schema is deployed on each. Needs the Citerank AI plugin connected.

page · per_page (max 100)

citerank_analyze_wp_page

Analyze one page on your connected WordPress site: the recommended schema type and why, extracted facts (phone, address, prices, FAQs, WooCommerce data) and author.

post_id (required)

citerank_deploy_schema

Publish JSON-LD and other agent files (llms.txt, robots rules, entity schema, WebMCP, A2A card) to your connected WordPress site, logged in your deployment history.

type (required) · payload (required) · post_id for per-page schema

citerank_get_wp_health

Check every deployment on your connected WordPress site: schema active, missing, modified or deleted per page, and the llms.txt, A2A, tool catalog and robots files.

none

Plans and limits

A tool call counts against your plan exactly like the same tool on the website, under the same limits. On top of those, each account has an allowance of MCP tool calls per rolling 24 hours: 10 on Free, 50 on Starter, 150 on Pro and 500 on Agency, with at most 10 calls a minute. A call over the allowance returns an error message your assistant can read, and costs nothing.

Every plan

citerank_analyze_url and citerank_generate_schema. On the free plan each call uses one of the 2 scans a month.

Starter and above

citerank_check_brand_citations and citerank_check_agentic_readiness (10 runs a month on Starter). On Starter every call above counts toward the plan's 50 runs a month. citerank_deploy_schema publishes to your connected site and is never counted.

Pro and above

citerank_simulate_agent_journey. Pro and Agency have no monthly cap on any of these tools.

When a call is outside your plan or over its limit, the tool returns an error naming the plan it needs.

Example prompt once connected: "Use citerank to audit example.com and tell me the top 3 things blocking it from being cited in AI Overviews."

REST API

Prefer plain HTTP? The same data is available through the Citerank REST API with API-key auth and per-plan rate limits.