Give your AI agent your AEO data

Connect Claude Code, Cursor, or any MCP client over OAuth. Your agent reads your visibility reports and citations across ChatGPT, Perplexity, Google AI Mode, Claude, Gemini, and Grok, runs new reports, and ships the fixes, all from your answer engine optimization (AEO) data, without leaving the tools you already work in.

Connect now Read the docs

OAuth 2.1 No key to paste Works with any MCP client

MCP server · Live

Connect Rank Prompt bash Live

# add the remote Rank Prompt MCP server (--scope user = available in every project)
claude mcp add --transport http --scope user rankprompt https://mcp.rankprompt.com/mcp

# in Claude Code, run /mcp to sign in over OAuth
# no API key to paste

Live at mcp.rankprompt.com/mcp

How it works

A remote server, not a local script

The Rank Prompt MCP server is hosted, so there is nothing to clone, install, or keep running. Your client connects to one URL over Streamable HTTP and authenticates in the browser.

In Claude Code you run /mcp and sign in over OAuth: no API key to copy, paste, or rotate. Clients without an OAuth flow can fall back to a pass-through key from the Developers page.

transport: Streamable HTTP (remote)
endpoint: mcp.rankprompt.com/mcp
auth: OAuth 2.1 (browser) or API key
clients: Claude Code, Cursor, Claude Desktop, any MCP client

Connect

Connect in 30 seconds

Pick your client, copy the snippet, then approve access in the browser.

Terminal bash

claude mcp add --transport http --scope user rankprompt https://mcp.rankprompt.com/mcp

Run the command, then /mcp to sign in over OAuth. No key to paste.

~/.cursor/mcp.json json

{
  "mcpServers": {
    "rankprompt": {
      "url": "https://mcp.rankprompt.com/mcp"
    }
  }
}

Add to your Cursor MCP config, then approve the OAuth prompt in the browser.

claude_desktop_config.json json

{
  "mcpServers": {
    "rankprompt": {
      "url": "https://mcp.rankprompt.com/mcp"
    }
  }
}

Settings, Developer, Edit Config. Paste, restart Claude Desktop, then approve access.

mcp.json json

{
  "mcpServers": {
    "rankprompt": {
      "url": "https://mcp.rankprompt.com/mcp"
    }
  }
}

Any MCP client with remote Streamable HTTP support. Point it at the URL and authenticate over OAuth, or paste a pass-through API key.

Tools

Twenty tools, split by what they touch

Read-only by default. The write tools are explicit, and the generic read, write, and delete tools reach any /v1 endpoint, so the agent can do anything the API can as it grows.

Discover

4 tools

Orient the agent before it touches your data.

rankprompt_prime
Start here: auth, the canonical loop, and your entitlement status.
rankprompt_schema
Find any /v1 endpoint by method, path, and summary.
rankprompt_skills_list
List the built-in playbooks (workflow, prompts, scheduling, seo-aeo).
rankprompt_skills_read
Read a playbook by name.

Read

8 tools

Pull your visibility data, read-only.

rankprompt_list_brands
List your brands, newest first.
rankprompt_get_brand_facts
A brand's researched facts (description, industry, value prop).
rankprompt_list_reports
A brand's reports, newest first.
rankprompt_get_report
Full report: prompts, per-platform results, and citations.
rankprompt_get_report_summary
Visibility score plus AI suggestions for a report.
rankprompt_list_citations
A brand's citations, filtered to the reachable ones.
rankprompt_get_job
Poll an async job until it completes.
rankprompt_read
Generic GET against any /v1 path.

Build

writes

4 tools

Set up brands, reports, and schedules.

rankprompt_research_brand
Research a brand's facts automatically (async, 1 credit).
rankprompt_create_report
Create a draft report and pick the engines.
rankprompt_add_prompts
Seed a report with prompts, or auto-generate them.
rankprompt_create_schedule
Re-measure a brand daily, weekly, or monthly.

Act

writes

4 tools

Trigger runs and write changes back.

rankprompt_run_report
Trigger an analysis run and get a job to poll.
rankprompt_audit_pages
SEO and AEO audit of a URL, list, or sitemap (1 credit each).
rankprompt_write
Generic POST or PATCH against any /v1 path.
rankprompt_delete
Generic DELETE against any /v1 path.

Agentic AEO

What you can ask your agent to do

Once connected, these are plain prompts. The agent picks the tools and runs the loop.

recipe 1

Pull my latest visibility report and implement the top 3 fixes.

tools

list_brands → get_report_summary → list_citations

recipe 2

Draft outreach for the highest-scoring domains that don't link to us yet.

tools

list_citations → read → write

recipe 3

Run a new report, wait for it, and open a PR with the changes.

tools

run_report → get_job → get_report_summary

Trust

Built to connect safely

Standard OAuth, scoped access, and encrypted credentials, on the same hardened infrastructure as the rest of the platform.

OAuth 2.1 + PKCE

Browser sign-in with PKCE (S256). No secrets in your shell history.

Dynamic registration

Clients register over DCR, then you approve access on a consent screen.

Scoped per-key access

Every call is bound to your key's scopes and brand restrictions.

Encrypted credentials

Stored credentials are encrypted at rest and revocable anytime.

Server-to-server

A hosted remote server. Browser CORS calls are blocked at the edge.

Plan

MCP access is included on every Starter plan and up, or on a standalone API plan. It is the same entitlement as the public /v1 API.

See plans

No OAuth flow in your client? Mint a pass-through key on the Developers page.

FAQ

MCP questions

Everything teams ask before they connect an agent. Email us if your question isn't covered.

What is the Model Context Protocol (MCP)?

MCP is an open standard for connecting AI assistants to external tools and data. The Rank Prompt MCP server exposes your AI visibility data, your brands, reports, and citations, as tools that any MCP client can call.

Do I have to paste an API key?

In Claude Code, no. Run /mcp and sign in through your browser over OAuth. Clients without a built-in OAuth flow can use a pass-through API key minted on the Developers page in your workspace.

Is the MCP server secure?

Yes. It uses OAuth 2.1 with PKCE, dynamic client registration, an explicit consent screen, scoped per-key access, encrypted credentials, and server-to-server delivery. Browser CORS calls are blocked at the edge.

How is the MCP server different from the public API?

Same data, different entrypoint. The /v1 REST API is for your backend to pull data; the MCP server lets your AI agent read and act on that data conversationally, inside the tools you already work in.

Which clients can connect?

Any MCP client that supports remote Streamable HTTP servers, including Claude Code, Cursor, and Claude Desktop. Point the client at https://mcp.rankprompt.com/mcp and authenticate over OAuth.

What can the agent actually do?

Read your brands, report summaries, and citations across ChatGPT, Perplexity, Google AI Mode, Claude, Gemini, and Grok, trigger new reports, and reach any /v1 endpoint through the generic read, write, and delete tools.

What plan do I need?

API access requires a Starter plan or higher, or a standalone API plan. It is the same entitlement as the public /v1 API.

Is the MCP server live?

Yes. It runs as a hosted remote server at mcp.rankprompt.com/mcp, alongside the public /v1 API and outbound webhooks.

Still have questions? Contact our team

MCP server · Live

7-day free trial · 50 free prompts

Bring Rank Prompt into your agent

Start a free trial, connect over OAuth, and let your AI agent work from your real AI visibility data.

Start free trial Connect now