> ## Documentation Index
> Fetch the complete documentation index at: https://docs.kime.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP Server

> Connect Kime to Claude, ChatGPT, Cursor, and other AI assistants so you can take action on your AI visibility — work your action plan and launch prompts, right from a chat.

Take action on your AI visibility without leaving your AI assistant. The Kime MCP server connects tools like Claude, ChatGPT, Cursor, and Claude Code — or any client that supports the Model Context Protocol (MCP) — directly to Kime, so you can work your action plan, create and update actions, and launch new prompts to track, all from a conversation.

You're connected to the same data the dashboard shows, so the context is always there when you need it. Dig into visibility, competitors, sentiment, and sources to understand the story behind an action — then act on it right away.

## What you can do

<CardGroup cols={2}>
  <Card title="Work your action plan" icon="list-check">
    Pull the actions Kime ranks by predicted impact, create new ones, and move them forward — your assistant confirms before it writes.
  </Card>

  <Card title="Launch & track prompts" icon="magnifying-glass">
    Add prompts in bulk to start tracking, then review how each one performs.
  </Card>

  <Card title="Track visibility & competitors" icon="chart-line">
    See how you show up across every AI model and where you stand against competitors on visibility, placement, and share of voice.
  </Card>

  <Card title="Understand AI perception" icon="brain">
    Explore sentiment, the keywords driving tone, and the sources AI models cite when they mention your brand.
  </Card>
</CardGroup>

## Setup guides

<Tabs>
  <Tab title="Claude">
    <Steps>
      <Step title="Open the Claude Desktop app">
        Launch Claude on your desktop.
      </Step>

      <Step title="Open Connectors">
        Go to **Customize** → **Connectors**.
      </Step>

      <Step title="Add a custom connector">
        Click the **+** sign to add a connector, then select **Add custom connector**.
      </Step>

      <Step title="Enter the Kime URL">
        Call it **Kime** and enter this URL: `https://mcp.kime.ai/mcp`
      </Step>

      <Step title="Connect">
        Press **Add**, and then **Connect**.
      </Step>
    </Steps>

    The first time you ask Claude about Kime data, it will prompt you to securely sign in.
  </Tab>

  <Tab title="ChatGPT">
    Custom MCP connectors in ChatGPT require **Developer mode** (available on Plus, Pro, Business, Enterprise, and Education plans). On Business and Enterprise workspaces, an admin may need to enable Developer mode first.

    <Steps>
      <Step title="Open ChatGPT">
        Open ChatGPT in your browser.
      </Step>

      <Step title="Open Advanced settings">
        Go to **Settings** → **Apps** → **Advanced settings**.
      </Step>

      <Step title="Turn on Developer mode">
        Turn on **Developer mode** and accept the warning.
      </Step>

      <Step title="Create the app">
        Click **Create app** (this only appears when Developer mode is on).
      </Step>

      <Step title="Enter the Kime URL">
        Name it **Kime**, paste this URL, and choose **OAuth** as the authentication method: `https://mcp.kime.ai/mcp`
      </Step>

      <Step title="Confirm the connection">
        Save the app, then use **Scan tools** or **Refresh** to confirm ChatGPT can reach Kime.
      </Step>

      <Step title="Enable the connector">
        In a new chat, open the **+** menu → **Developer mode**, and enable your Kime connector.
      </Step>
    </Steps>

    The first time ChatGPT uses a Kime tool, it will prompt you to securely sign in.
  </Tab>

  <Tab title="Other clients">
    Kime works with any MCP-compatible client over Streamable HTTP. Point it at the connection URL and sign in via OAuth when prompted.

    <CodeGroup>
      ```bash Claude Code theme={null}
      claude mcp add kime --transport http https://mcp.kime.ai/mcp
      ```

      ```json Cursor (mcp.json) theme={null}
      {
        "mcpServers": {
          "kime": {
            "url": "https://mcp.kime.ai/mcp"
          }
        }
      }
      ```

      ```json VS Code (.vscode/mcp.json) theme={null}
      {
        "servers": {
          "kime": {
            "type": "http",
            "url": "https://mcp.kime.ai/mcp"
          }
        }
      }
      ```
    </CodeGroup>

    In **Cursor**, you can also open **Settings → Tools & Integrations → MCP**, add a custom MCP server with the Kime URL, and select Streamable HTTP.
  </Tab>
</Tabs>

For full setup steps and troubleshooting across every client, see [Connect AI assistants](/administration/integrations).

## Example questions

Ask in plain language. Here are some prompts to get you started:

**Visibility & performance**

* "How has our visibility changed over the last 30 days?"
* "Which AI models mention us the most?"
* "Compare our visibility in ChatGPT versus Gemini this month."

**Competitors**

* "Which competitors have the highest share of voice?"
* "Where do competitors show up but we don't?"
* "Compare our placement against \[competitor] over the last 14 days."

**AI perception**

* "What's our sentiment score across AI models right now?"
* "Break our sentiment down by factor and show the strongest negative excerpts."
* "Which keywords drive negative sentiment about our brand?"
* "Summarize how AI models currently describe us."

**Sources & citations**

* "Which domains get cited most when our brand comes up?"
* "What sources are behind the most positive mentions of us?"

**Take action**

* "What should we work on next to improve our AI visibility?"
* "List our open actions and create one to publish a comparison page versus \[competitor]."
* "Mark the \[competitor] comparison action as in progress and add a note with the angle."
* "Add these 10 prompts and start tracking them."

## Authentication

Kime uses **OAuth** — there's nothing to paste or store. The first time your assistant uses a Kime tool, it opens a secure sign-in window where you log in and approve access. After that, your session persists across conversations.

Access always respects your Kime account: an assistant can only see the organizations and workspaces you can, and acts within your existing permissions. The consent screen identifies which application is requesting access, and you can switch accounts if needed.

## Read and write access

Most tools are **read-only**. Visibility, sentiment, sources, competitor benchmarks, prompts, and actions all pull existing data without changing anything.

A smaller set of tools help you act on what you find:

* **Work your actions** — pull the prioritized action plan Kime surfaces, create new actions, update their status, and add comments as you make progress.
* **Create prompts** — add prompts in bulk and start tracking them across your models.

Write tools are clearly flagged, so your client asks for confirmation before anything runs. You grant the matching permissions when you connect, and changes always stay within your existing Kime access.

## Supported platforms

The Kime MCP server works with any client that supports the Model Context Protocol, including Claude, ChatGPT, Cursor, Claude Code, VS Code (GitHub Copilot), and Windsurf.

## Requirements

A Kime account with at least one active workspace, and an MCP-compatible AI tool.