Ask your AI visibility data in plain language.
Connect Rankwize to Claude Desktop, Cursor, Claude Code, or any MCP-compatible client. Ask a question; your assistant queries your citation data, prompt status, and effectiveness scores, and answers from live data. No exports, no manual filtering, no leaving your workflow.
MCP is Anthropic's open standard for letting AI assistants call tools on your behalf. Included on Pro and above. Read-only by default.
What you ask. What it returns.
Three patterns from teams using Rankwize MCP today. Each question maps to one or two tool calls; your assistant handles the orchestration.
SEO LEAD
"Which of my top organic queries from last month are not yet cited by any AI platform?"
Calls
get_top_queries list_prompts Cross-references GSC queries against your tracked prompts, surfaces high-impression queries with invisible status on ChatGPT, Perplexity, and Google AI Overviews, ranked by impression volume.
CONTENT DIRECTOR
"For my top five fix types over the last 90 days, what is the citation rate and which are most underperforming?"
Calls
get_topic_effectiveness Returns citation rate per fix type with run counts. Your assistant ranks the underperformers so the next content sprint targets where the citation needle actually moves.
AGENCY OPERATOR
"Summarize last week's runs across all monitored prompts, grouped by platform, and post it to my Monday standup channel."
Calls
get_run_summary Pulls last week's run counts, citations, and failure rates per platform. Paired with a Slack MCP server, lands the standup summary in channel without anyone opening Rankwize.
Ten tools. Read-only by default.
Eight tools read your monitoring, diagnosis, and search data. Two write tools (
create_prompt and
trigger_run) require explicit
write scope on the API key. A read-only key
can never mutate your workspace.
Track plans expose the monitoring tools: prompts, runs, citations, content library. +Diagnose adds the diagnosis tools: fix types, recommendations, topic effectiveness.
list_prompts Tracked intents with current citation status per platform
get_prompt_detail Factor scores, citation history, matched page, current recommendation. Diagnosis fields require +Diagnose
get_content_library Indexed content pages with fingerprint data
get_topic_effectiveness Effectiveness scores by topic, fix type, and platform. +Diagnose only
get_run_summary Results from recent monitoring batch runs
get_top_queries Top organic search queries from Google Search Console. Requires GSC integration.
get_top_landing_pages Top landing pages by organic search performance from GSC, with a tracked-in-Rankwize flag.
get_ai_referral_traffic AI-platform referral traffic from Google Analytics 4. Requires GA4 integration.
create_prompt Add a new prompt intent to your tracking queue
trigger_run Enqueue an on-demand monitoring batch
Tool: list_prompts
Authorization: Bearer rwz_live_xxxx
→ {
"prompts": [
{
"id": "pi_01jxxxxxxx",
"intent": "best ai visibility tool for b2b saas",
"platforms": {
"chatgpt": "cited",
"perplexity": "invisible",
"google_ai_overviews": "mentioned"
},
"priority_score": 82,
"fix_type": "Add_Evidence"
}
],
"total": 94
} Setup in under five minutes.
Generate a key
Settings → API Keys → New key. Assign read or write scope. Label per agent or workflow so you can revoke selectively.
Add the server to your client
In Claude Desktop, use Customize → Add custom connector to connect via OAuth — no config file needed. For Cursor, Claude Code, and others, paste the server URL and pass your API key as a Bearer token.
Ask
Your assistant fetches the tool schemas on connect and routes your natural-language questions to the right tool calls.
{
"mcpServers": {
"rankwize": {
"transport": "sse",
"url": "https://api.rankwize.io/mcp/sse",
"headers": {
"Authorization": "Bearer rwz_live_YOUR_KEY"
}
}
}
} Tested with
Read-only by default. Auditable by design.
The MCP Server is a pull surface, not a push surface. You decide which keys exist, what they can do, and when they stop working.
Per-tenant keys, separately revocable
Each key is scoped to one workspace and one set of permissions. Label per agent or user. Revoke one without touching the others. Keys are hashed at rest; the raw value is shown once and cannot be recovered.
Write scope is opt-in
The two write tools require the
write scope to be checked at key creation. A
read-only key cannot create prompts or trigger runs even if asked.
Audit log on every call
Every tool call is recorded with the tool name, key label, success or failure, and an argument hash. Ninety-day retention. Raw arguments and response payloads are never logged.
Rate-limited and quota-bounded
Per-minute throttles protect the service from runaway clients. Monthly MCP units cap cumulative usage per workspace. A buggy agent cannot exhaust your account silently.
Included on Pro and above.
The MCP Server ships with every paid plan from Pro upward, on both Track and Diagnose. No separate purchase. Each workspace gets a monthly MCP unit allowance, shared across all keys in that workspace, that resets on the first of each month (UTC).
Plan
Track
+ Diagnose
Pro
10,000 units/mo
25,000 units/mo
Business
50,000 units/mo
75,000 units/mo
Scale
150,000 units/mo
300,000 units/mo
Most read calls cost 1 unit. Heavier reads and
create_prompt cost 5.
trigger_run costs 100 (one full citation check
across your monitored platforms). Free and Watch plans do not include MCP.
Included on Pro and above. Track and Diagnose.