---
title: "Getting Started"
slug: "aia-mcp-getting-started"
description: "Set up the Kentik AI Advisor MCP Server, including the required endpoints, headers, and how to get your Kentik API Token."
tags: ["Kentik API", "MCP Server", "AI Agent"]
status: "update"
updated: 2026-11-10T03:42:57Z
published: 2026-11-10T03:42:57Z
canonical: "kb.kentik.com/aia-mcp-getting-started"
---

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

# Getting Started

This article describes how to set up Kentik’s [**AI Advisor MCP Server**](/v1/docs/ai-advisor-mcp-server).

> [!TIP]
> 🏃‍♀️ **Tip:** Already familiar with the Model Context Protocol (MCP)? Go to [**Configuration Examples**](/v1/docs/aia-mcp-configuration-examples) and follow the steps for your preferred agent.

## **Prerequisites**

- An active Kentik account with [**AI Advisor enabled**](/v1/docs/ai-advisor#about-ai-advisor).
- Kentik API credentials (email and [**API token**](/v1/docs/aia-mcp-getting-started#get-your-api-token)), if you use API token authentication.
- An MCP-compatible AI agent or development tool, such as Claude Desktop or GitHub Copilot (see [**Configuration Examples**](/v1/docs/aia-mcp-configuration-examples)).
- Your MCP client must use version 2026-07-28 or 2025-11-25. Clients using earlier versions, such as 2024-11-05, time out when connecting to AI Advisor.

## **Get Your API** **Token**

To authenticate with an API token, copy the token from your user profile.

> [!NOTE]
> **Note**: This step does not apply if you authenticate with OAuth.

1. In the Kentik portal, navigate to your **User Profile** (top-right menu).
2. Select the **Authentication** tab.
3. Copy your **API Token**.

![User profile settings with Single Sign-on information and API token section displayed.](https://cdn.us.document360.io/082e25b5-afce-42d4-8f47-70bd3f1d02b7/Images/Documentation/user-profile-api-token.png?sv=2026-02-06&spr=https&st=2026-11-10T16%3A08%3A32Z&se=2026-11-10T17%3A08%3A32Z&sr=c&sp=r&sig=e%2B4XVmoeuc0H4lDNVwGlXauB6ll6atsZC33iVIeZ8vc%3D)

## Connect the MCP Server

To connect a client, paste a JSON snippet or run a terminal command, depending on the client:

1. Select the [**Configuration Example**](/v1/docs/aia-mcp-configuration-examples) for your agent and authentication method (OAuth or API token).
2. For API token authentication, replace the [**endpoint**](/v1/docs/aia-mcp-getting-started#endpoints) and [**header**](/v1/docs/aia-mcp-getting-started#required-headers) values in the example with the Endpoint URL for your region and the Required Headers.
3. Follow the [**instructions**](/v1/docs/aia-mcp-configuration-examples) to apply the configuration in the AI agent or development tool.
4. Start a [**conversation**](/v1/docs/aia-mcp-getting-started#usage) with AI Advisor. Remember to mention "Kentik" in your prompts.

> [!NOTE]
> **Note**: If you’re using a custom AI agent framework, see [**Custom Applications**](/v1/docs/aia-mcp-configuration-examples#custom-applications) and [**MCP Payload Reference**](/v1/docs/aia-mcp-payload) for additional details.

### Endpoints

Choose the URL for your region or cluster.

| Region | Endpoint URL |
| --- | --- |
| **US Cluster** | `https://api.kentik.com/mcp` |
| **EU Cluster** | `https://api.kentik.eu/mcp` |

### Required Headers for API Token Authentication

OAuth does not require these headers.

| Header Name | Value |
| --- | --- |
| `X-CH-Auth-Email` | <your-email@company.com> |
| `X-CH-Auth-API-Token` | <your-api-token> |

## AI Advisor MCP Usage

Here are some tips on interacting with the AI Advisor MCP Server via your AI agent.

### Effective Prompting

Mentioning "**Kentik**" in your initial prompts helps ensure the external AI agent invokes the [**Kentik MCP tools**](/v1/docs/aia-mcp-tools). Otherwise, the agent might return a generic or incorrect response.

**Examples**: `"Show me vmx devices from Kentik"` or `"What is the status of my Kentik devices?"`.

### **Real-Time Progress**

While AI Advisor processes your query, you see real-time progress updates in supported MCP clients (like Claude Desktop and GitHub Copilot). These notifications show the AI’s reasoning process:

- `🔍 Created session...` - Initial session created
- `⏱️ [2s] Analyzing device health metrics...` - Processing updates
- `⏱️ [4s] Checking interface utilization...` - Continued progress
- `✅ Completed` - Final answer ready

Next, see [**Configuration Examples**](/v1/docs/aia-mcp-configuration-examples) for the AI Advisor MCP Server.