MCP server

The AI visibility MCP server for Claude Code, Cursor and Codex

Connect Claude Code, Cursor, Codex and other MCP clients to your AI search data, then ask about visibility, citations and competitors in plain English. Every answer comes from the same numbers as your dashboard.

Get started Read the setup guide

Server URLhttps://mcp.geoptie.com/mcp

Works with: Claude Code · Cursor · Codex · Any remote MCP client

48
tools to read and act
120/min
requests, no monthly cap
1 key
shared with the REST API
$99
per month, Professional plan

Trusted by marketing teams and agencies

What is an AI visibility MCP server?

An MCP server lets an AI assistant call a software product directly, through the Model Context Protocol. Geoptie's MCP server gives assistants like Claude Code, Cursor and Codex the same AI visibility data your team sees in the dashboard: how often each engine mentions your brand, the full answers, the sources they cite, and how you compare with competitors.

24 of its 48 tools act as well as read, so the assistant can also start tracking prompts, work through recommendations, run page audits and draft content, in the conversation you are already having.

Connect in one command

The server runs at mcp.geoptie.com over Streamable HTTP and takes your API key as a bearer token. Create the key in the dashboard, then add the server to your client.

1
Create a key
In the dashboard under API keys. The same key works for the REST API.
2
Add the server
One command in Claude Code, or a few lines of config in Cursor and Codex.
3
Ask in plain English
Start with "List my Geoptie brands", then ask about visibility, citations or competitors.

terminal

claude mcp add --transport http geoptie https://mcp.geoptie.com/mcp \
  --header "Authorization: Bearer $GEOPTIE_API_KEY"

.cursor/mcp.json

{
  "mcpServers": {
    "geoptie": {
      "url": "https://mcp.geoptie.com/mcp",
      "headers": {
        "Authorization": "Bearer ${env:GEOPTIE_API_KEY}"
      }
    }
  }
}

~/.codex/config.toml

[mcp_servers.geoptie]
url = "https://mcp.geoptie.com/mcp"
bearer_token_env_var = "GEOPTIE_API_KEY"

stdio via mcp-remote

{
  "mcpServers": {
    "geoptie": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://mcp.geoptie.com/mcp",
        "--header", "Authorization: Bearer ${GEOPTIE_API_KEY}"
      ]
    }
  }
}

Example prompts

Which of our prompts lost visibility this month, and on which engines?
Which pages do AI engines cite most when they answer about us?
Compare our share of voice against our competitors over the last 30 days.
What are the open recommendations for this brand, and which look most worthwhile?
Start tracking these five prompts and file them under a new topic called Pricing.
Take the top recommendation and write an article for it.

Everything the dashboard covers

Each tool is one call to the Geoptie API, so the assistant sees the same metrics, answers and citations you see in the dashboard.

Visibility and share of voice

Visibility, position and share of voice across the engines each brand tracks, with the previous period alongside so a number becomes a direction.

get_brand_visibility
list_competitors
get_competitor_trends

Answers and citations

The full text each engine returned, the brands it named, and the pages and domains it cited, ranked by how often they appear.

list_answers
list_prompt_mentions
list_cited_urls
list_cited_domains
get_citation_trend

Competitors

Your brand against its competitors on the same prompts, plus tools to track, edit or ignore competitors without opening the dashboard.

list_tracked_competitors
track_competitors
ignore_competitors

Prompts and topics

Start or stop tracking prompts and organise them into topics from the conversation. Re-running a request never creates duplicates.

list_prompts
track_prompts
untrack_prompt
create_topic
update_prompt

Recommendations

Work through the open recommendation queue with the evidence behind each item, then complete, dismiss or restore them.

list_recommendations
generate_recommendations
update_recommendation

Audits and content

Run a page audit, generate a content brief, and write the article from that brief, all from the assistant.

create_audit_report
create_generation
generate_draft
create_optimization

Built to be safe to hand an assistant

Read and write, clearly labelled

Of the 48 tools, 24 read and 24 change data. Every tool carries read-only or destructive hints, so clients can ask before acting.

Brand-scoped keys

A key can cover the whole workspace or only the brands you choose. A scoped key behaves as though the other brands do not exist.

Limits that stop runaway loops

120 requests per minute per key and no monthly quota. Tools that run a model are limited to a few calls per minute and two at a time.

Same numbers as the dashboard

Precomputed metrics, not raw rows to reassemble, so the assistant's answer matches what your team sees.

Included on Professional and Enterprise

AI visibility MCP server FAQ

How is the MCP server different from the API?

The REST API serves code: a scheduled job that copies data into a warehouse, a client dashboard, an alert. The MCP server serves AI assistants, which call it during a conversation when you ask a question. Both use the same key, the same data and the same limits, so many teams use the API for reporting and the MCP server for day-to-day questions.

Which AI assistants can connect to Geoptie's MCP server?

Claude Code, Cursor and Codex have documented setups, and any client that supports remote MCP servers with custom headers connects the same way. Clients that only run local servers over stdio can connect through the mcp-remote proxy. The server takes an API key in a header rather than a sign-in flow.

Do I need a developer to set it up?

No. Create a key in the dashboard, then run one command in Claude Code or paste a few lines of config into Cursor or Codex. Asking "List my Geoptie brands" confirms the connection in a couple of minutes.

Which plans include the MCP server?

Professional ($99 a month) and Enterprise ($399 a month). The MCP server and the REST API share one key, created in the dashboard under API keys. Starter does not include API or MCP access.

Can the assistant change my data?

Yes, if you let it. 24 of the 48 tools create, update or delete things such as tracked prompts, topics, competitors and recommendations, and every tool is labelled read-only or destructive so your client can ask before acting. A brand-scoped key limits the assistant to the brands you choose.

Are there usage limits?

Each key can make 120 requests per minute, with no monthly request quota. Tools that run a model, such as audits, recommendations and content generation, are limited to a few calls per minute and two at a time. These limits apply to the API and MCP only, never to your dashboard usage.

Is the data the same as in the dashboard?

Yes. The server returns the same precomputed visibility, competitor and citation metrics the dashboard shows, plus the full text of the answers the engines gave.

Bring Geoptie into the tools you already use

Start on Professional, create a key, and your assistant can answer AI visibility questions in a couple of minutes.

Get started