---
title: "TARGIT AI Connect MCP Setup Guide"
canonical: "https://support.targit.com/space/TT/573767688/TARGIT%20AI%20Connect%20MCP%20Setup%20Guide"
format: markdown
---
The TARGIT AI Connect MCP server allows AI assistants such as Claude, ChatGPT, and Microsoft Copilot to securely access your TARGIT Cloud environment and query metadata, cubes, dimensions, and data.

It requires a license to the TARGIT AI Connect Essentials module. Contact your sales rep for more information.

For a quick introduction to what you can do with the TARGIT AI Connect MCP feature, watch this feature video:

### **Prerequisites**

Before you begin, ensure that:

- You have access to a TARGIT Cloud environment.
- Your environment is reachable via HTTPS.
- You have obtained a valid license for the TARGIT AI Connect module.
- You have a valid TARGIT user account with proper permissions.
- Your MCP endpoint URL is available.

The *MCP Data Access *rights setting with the *TARGIT AI Connect *license module in the TARGIT Management client:

![image-20261008-062636.png](media://9724f6da-ed9b-41d0-bf90-e5a4bd234afc)


### **MCP Endpoint Format**

Replace [tenantname] with your TARGIT tenant name:

- https://[tenantname].targit.cloud/server/request-engine/mcp/

Example:

- [https://customer.targit.cloud/server/request-engine/mcp/](https://contoso.targit.cloud/server/request-engine/mcp/)

---

## **Available MCP Tools**

When connecting for the first time, your AI client will request permission to use six TARGIT tools.

We recommend selecting **Always Allow** for all tools to avoid repeated approval prompts.

|  |  |
| --- | --- |
| **Tool** | **Description** |
| Get catalog | Retrieves the catalog structure and available content |
| Get cube | Retrieves metadata for a specific cube |
| List connections | Lists available data source connections |
| List cubes | Lists all available cubes |
| List members | Lists available members within a dimension |
| Run request | Executes analytical requests against TARGIT |

After approval, your AI assistant can use these tools to discover available data and answer analytical questions.

---

## **Connecting to Claude Desktop**

Note that connecting an LLM to TARGIT’s MCP Server requires external knowledge of adding MCP functionality to that LLM. The following guidelines should be regarded as examples only.

#### **Step 1: Open Claude Desktop Settings**

1. Open Claude Desktop.
2. Click **Settings**.
3. Navigate to **Connectors** or **MCP Servers**.

![image-20261007-064902.png](media://b7aa4aac-a782-46dd-9eb2-83f75d227433)


#### **Step 2: Add a New MCP Server**

Choose **Add custom connector **and enter:

- **Name: **TARGIT
- **URL: **https://[tenantname].targit.cloud/server/request-engine/mcp/

![image-20261007-065010.png](media://ac368135-2f13-4f50-9010-a0c2c2d6aedb)


#### **Step 3: Sign In**

Claude Desktop will redirect you to the TARGIT authentication page.

1. Sign in using your TARGIT credentials.
2. Complete any required multi-factor authentication.

![image-20261007-065136.png](media://6126f5fa-e0ff-4790-bcac-6ab034eb20e0)


#### **Step 4: Approve Tools**

You will be presented with the six available tools.

Select **Always Allow **for:

- Get catalog
- Get cube
- List connections
- List cubes
- List members
- Run request

![image-20261007-065309.png](media://9d055c96-985d-4a06-bd06-2f036a566368)


#### **Step 5: Verify the Connection**

Try a prompt such as:

- What cubes are available in TARGIT?

If the server is connected successfully, Claude will retrieve information from TARGIT.

![image-20261007-065426.png](media://93d26890-5a62-447b-ae16-463c0fb0bc3c)


---

## **Connecting to ChatGPT**

Note that connecting an LLM to TARGIT’s MCP Server requires external knowledge of adding MCP functionality to that LLM. The following guidelines should be regarded as examples only.

#### **Step 1: Open ChatGPT**

Open ChatGPT and navigate to:

**Settings → Connectors**

(or the MCP/Connectors section depending on your ChatGPT version).

#### **Step 2: Add Connector**

Select:

Add Connector

or

Add MCP Server

#### **Step 3: Enter Server Details**

**Name**

TARGIT

**Server URL**

https://[tenantname].targit.cloud/server/request-engine/mcp/

#### **Step 4: Authenticate**

When prompted:

1. Sign in to your TARGIT environment.
2. Authorize ChatGPT to access TARGIT through MCP.

#### **Step 5: Approve Tools**

Grant access to all six tools.

Recommended setting:

Always Allow

for each tool.

#### **Step 6: Test the Connection**

Example prompts:

List all cubes available in TARGIT.

Show available dimensions in the Sales cube.

What measures are available in the Finance cube?

---

## **Connecting to Microsoft Copilot**

Note that connecting an LLM to TARGIT’s MCP Server requires external knowledge of adding MCP functionality to that LLM. The following guidelines should be regarded as examples only.

#### **Step 1: Open Microsoft Copilot Studio**

Open Microsoft Copilot Studio, [https://copilotstudio.microsoft.com/](https://copilotstudio.microsoft.com/), and navigate to the **Agents** area.

#### **Step 2: Add New agent**

Add a Standard agent.

![image-20261007-084719.png](media://0ca6d93e-21f5-40c6-b5af-a138ae05ef22)

Name the agent, any name will work:

![image-20261007-084806.png](media://e1a8a7f4-7a7a-407f-a7bd-85753331ecb1)

#### **Step 3: Add tool to the Agent**

Open the **Tools** tab and select **Add a tool**.

Select the **Model Context Protocol ** tool and provide Server name (any name will work), Server description and Server URL for the tool.

**Server URL: **https://[tenant name].targit.cloud/server/request-engine/mcp/

Replace [tenant name] with the name of your TARGIT tenant.

![image-20261007-085631.png](media://d49121ed-46d2-4b34-b8a8-286ce5ba1e70)


#### **Step 4: Authentication**

Select the Authentication method that will work for your setup and click **Create**.

![image-20261007-085858.png](media://c2d11885-abcf-4ed8-b202-85578740db51)


Depending on your organization's configuration, you may be redirected to your identity provider for authentication.

#### **Step 5: Create new connection**

Select **Create new connection** to get connected to the MCP Server.

#### **Step 6: Verify Tools**

Once connected, you should have access to the tools:

- List connections
- Get catalog
- Run request
- List cubes
- List members
- Get cube

![image-20261007-090441.png](media://0c62e67a-1b7e-4740-ae39-a6c479067945)

#### **Step 7: Verify Access**

In the **Test your agent **area, try a simple query:

- *List all available cubes in TARGIT*

or

- *What information is available in the Sales cube?*

![image-20261007-090739.png](media://7473d0c1-3cbc-440c-a1b6-543c6ea09277)

#### **Step 8: Publish the Agent**

Press **Publish** to publish the Agent.

![image-20261007-091051.png](media://5aff63ac-9a41-4d47-a461-ed180bb04c6b)


---

## **Example Questions**

Once connected, you can ask questions such as:

#### **Discovery**

What cubes are available?

What dimensions exist in the Sales cube?

List available measures in the Finance cube.

#### **Analysis**

Show revenue by month for the current year.

Compare sales by region.

What were the top 10 customers by revenue last quarter?

#### **Metadata Exploration**

Explain the Gross Margin measure.

What members exist in the Product Category dimension?

Describe the structure of the Sales cube.

`

---

## **Troubleshooting**

#### **I cannot connect to the MCP server**

Verify:

- The tenant name in the URL is correct.
- The URL is reachable from your network.
- Your TARGIT account is active.
- Authentication completed successfully.

#### **I am repeatedly asked for permission**

Set each tool permission to:

Always Allow

instead of approving each request individually.

#### **No cubes or data are visible**

Verify that:

- Your TARGIT user has access to the required cubes.
- The connection has completed successfully.
- The AI assistant has been granted access to all six tools.

---

## **Security Note**

The AI client only has access to data that your TARGIT user account is authorized to access. All requests are executed using your existing TARGIT permissions and security model.