Most AI-visibility tools hand you a dashboard. The Cituna MCP server hands the work to your agent. Connect it once and Claude can read how the AI engines cite your brand, tell you which buyer prompts you are losing and why, then do something about it: change what you track, work your fix queue and queue the article AutoSEO writes next, without you leaving the chat or exporting a thing.
What it does
MCP, the Model Context Protocol, is an open standard for connecting AI assistants to live data and tools. Our server implements it, so any MCP client can query your workspace directly. Point it at Claude and you can ask about the seven engines we track, ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews and Google AI Mode (we do not track Microsoft Copilot), with the tracking grid refreshed daily across all seven. To call the same data from your own code, the developer docs cover the API and the plans that include it.
What that looks like in practice:
- Your visibility, on demand. “What changed in my AI visibility this week, and which engines dropped us?”
- The receipts. “What did Perplexity actually answer for ‘best CRM for small business’, and who did it cite instead of us?”
- The fix queue. “List the open gaps for acme.com and walk me through the schema one.”
- Search Console, joined up. “How did acme.com do in Search over the last 28 days, broken down by country then device?”
- Act on it (Pro). “Add the three questions we lost this week to our tracked prompts, queue an article for the biggest one, and re-scan.”
Why it matters:
The 26 tools
26 tools ship today: 20 reads and six writes. Reads work from $39 (Starter and up), with one exception: list_keywords reads the keyword dataset that is part of Pro. The six write actions, which change your workspace, need Pro or higher.
| Tool | What Claude does with it | Plan |
|---|---|---|
whoami | Confirm the connection and see your account, workspace, plan and usage. | From $39 |
list_brands | List the domains tracked in your workspace. | From $39 |
get_visibility | Read today’s tracking grid: which of the seven engines cited you, for which prompts, and at what position. | From $39 |
get_engine_answers | Pull the receipts: what each engine actually answered for a prompt, and which brands it cited instead. | From $39 |
list_audits | List recent AI-visibility audits, scores and open-gap counts, newest first. | From $39 |
get_audit | Open one audit in full: per-engine citations and the prompt × engine matrix. | From $39 |
list_gaps | Read a domain’s fix queue, each gap and its concrete fix. | From $39 |
gsc_status | Check whether Google Search Console is connected and which properties are available. | From $39 |
gsc_overview | Summarise Search Console clicks, impressions, CTR and average position for a domain. | From $39 |
gsc_query | Run any Search Analytics query, choose dimensions, filters and a date range. | From $39 |
gsc_striking_distance | List the queries ranking just off page one, ordered by how much a realistic move would unlock. | From $39 |
gsc_zero_click_pages | Find pages Google shows and nobody clicks, split by cause, because each cause needs a different fix. | From $39 |
gsc_query_mix | Group the demand you earn by intent: brand, competitor, commercial, how-to and category searches. | From $39 |
gsc_coverage_gap | List sitemap pages Google has never shown anyone, grouped by section. | From $39 |
gsc_cannibalisation | Find queries where two of your own pages compete, and which one Google favours. | From $39 |
gsc_ai_searches | Pick out the searches worded like AI prompts rather than keywords, as a floor on AI use. | From $39 |
list_citation_sources | See the third-party pages the engines cite for your prompts, and whether you appear on them. | From $39 |
competitor_discovery | Find rivals the engines name, or people search for, that you are not tracking yet. | From $39 |
list_keywords | Read keywords with real search demand and the stage of the page behind each, filter for demand nothing is written for yet. | Pro |
list_content_queue | See what AutoSEO already has drafted or in flight, so nothing gets queued twice. | From $39 |
queue_article | Queue a keyword so AutoSEO drafts the page on its next run. | Pro |
mark_article_published | Tell Cituna a page for a keyword is live, so it stops suggesting the topic. | Pro |
set_gap_status | Move a gap along your queue, mark it todo, doing or done. | Pro |
run_scan | Run a fresh AI-visibility scan for a URL (uses one scan from your quota). | Pro |
set_prompts | Add, remove, enable or disable the buyer questions the seven engines are asked every day. | Pro |
set_outreach_status | Mark a citation source planned, pitched, placed or declined, so the outreach list survives between sessions. | Pro |
Reads work on Starter and up (list_keywords needs Pro's keyword data); the six write tools (run_scan, set_prompts, set_gap_status, queue_article, mark_article_published, set_outreach_status) require Pro or higher. The free trial does not include MCP access. Tool set current as of September 26, 2026.
Connect it to Claude
Three steps, about a minute. Nothing to install, no config file and no key to copy: you approve the connection in your browser, decide there whether Claude may make changes as well as read, and can revoke it from Integrations whenever you want.
Add the connector
In Claude, open Customize → Connectors, click Add, then Add custom connector. Paste this URL:
connector URLhttps://mcp.cituna.com/mcpApprove access
Claude sends you to Cituna to sign in, then shows what it is asking for. Leave the write box ticked to let it run scans and work your queue, or untick it for read-only. Revoke whenever you like from Integrations. Needs Starter ($39) or higher.
Ask your first question
That is it, no terminal and nothing to install. Good openers:
- “What changed in my AI visibility this week?”
- “Which buyer prompts am I losing, and who gets cited instead?”
- “Which Search Console queries lost clicks this month?”
Prefer to run it on your own machine?
The server is also on npm, which is the better fit for scripts and CI. Generate a key first (Integrations → Claude / MCP access → Generate token), then:
claude mcp add cituna --scope user \
--env CITUNA_API_KEY=YOUR_TOKEN \
-- npx -y cituna-mcpAny other MCP client works too: the command is npx -y cituna-mcp with CITUNA_API_KEY in its environment. This path needs Node 20+ on your machine; the connector above does not.
Which plan you need
The MCP server is included on every paid plan, there is no separate add-on, and it is not part of the 3-day free trial. On Starter ($39), the read tools let Claude query your tracking grid, audits, gaps, content queue and Search Console data. Pro ($119) adds the keyword dataset and the six write tools, so Claude can start a fresh scan, change the prompts you track, work your fix queue, queue or mark up articles and log outreach for you; Max ($399) raises the scan, data and MCP allowances; and a custom Enterprise plan lifts them further with invoiced billing and priority onboarding. See the full pricing for the per-plan limits.
Drafted with AI assistance from our own research and Search Console data, and reviewed by Rahul A before publishing. Rules and prices change; check the linked official source before you act.