> ## Knowledge Base Index
> Fetch the complete knowledge base index at: https://kite.skailama.app/sitemap.xml
> Use this file to discover available pages before exploring further.
> Pure-Markdown content can be obtained by appending a '.md' suffix to the content URLs listed in the sitemap (without the trailing slash).

# Connect Kite to AI tools (Claude, Codex) via MCP

1. In the Kite app, navigate to the **Settings** section.
2. Click on **Connect Kite to AI Tools via MCP** and choose your preferred AI tool method (e.g., Claude).
3. Copy the generated connector URL.
4. Open the Claude app, navigate to **Settings**, select **Connect**, and choose to add a **Custom Connector**.
5. Give the connector a recognizable name (e.g., "Kite Analytics") and hit **Continue**.
6. Leave the remaining configuration settings as they are and click **Add**.
7. After adding the connector, click **Connect** to authenticate the integration. This will automatically open a new tab in your web browser.
8. Complete the authentication by clicking **Allow** in the browser tab.
9. Return to your Claude app, where you will now see the Kite tools available. Adjust your data reading permissions (e.g., set to "Always allow" or configure it so the tool requests permission before reading data).
10. Start a new chat and ask strategic questions like, "Which campaign brought in the most revenue?" to get tailored advice on your discount strategies based on your store's full context.

**Technical Note/Limitation:**

* **Platform Requirement:** This integration requires an AI application that explicitly supports the Model Context Protocol (MCP) standards.
* **Data Access:** The AI tool analyzes your store's data by reading your Kite analytics. Ensure your permission toggles in the AI app reflect your security preferences for data access.

**Troubleshooting/Checklist:**

* **Authentication tab doesn't open:** Check if your browser or operating system is blocking pop-ups or external link requests from the AI app.
* **AI cannot access data:** Verify that the tool permissions in the AI app's settings are set to allow data reading, rather than blocking the connection.
* **Missing connector option:** Ensure your AI app is updated to a version that supports adding custom MCP integrations.