Just as the Metric Insights Concierge service can connect to external BI tools to answer questions about corresponding data, other applications can also connect to Concierge. Metric Insights MCP server is a Model Context Protocol server that's built into the Metric Insights application. It exposes Concierge tools to external AI clients, including Claude, ChatGPT, Slack, Microsoft Teams / Copilot, and others.
Please use this guide to integrate Concierge into the AI application you're currently using.
PREREQUISITES:
The Concierge service has to be enabled and configured on the MI instance.
In this article:
- Collect Information from Metric Insights
- Configure AI Application
NOTE: If your organization's server is protected by a VPN, connectivity to the MI MCP server will be disrupted.
1. Collect Information from Metric Insights
To proceed with integration, you'll need the MCP endpoint, Client ID, and Client Secret.
1.1. Copy Metric Insights MCP Endpoint
The AI application can request an MCP Endpoint URL. Here's a list of MI MCP endpoints:
| Endpoint Purpose | Endpoint URL |
|---|---|
| OAuth Endpoint | https://<mi-host>/mcp |
| Authorize | https://<mi-host>/mcp/authorize |
| Token | https://<mi-host>/mcp/token |
| Callback (register on external App) | https://<mi-host>/mcp/auth/callback |
Instead of <mi-host>, insert your Metric Insights instance name.
1.2. Collect Client ID and Client Secret
In Metric Insights instance, access Admin > System > External Applications. If you already have a Concierge External Application, click on it to open the Editor. Otherwise, create an External Application for Concierge according to the guide below.
1.2.1. Set Basics
- [+ New External Application]
- Name: Define the name of an External Application.
- Auth Type: Define how the application will handle authentication:
- OAuth: This is the recommended option. In this case, MI acts as a standard OAuth2 provider.
- Token: If you select this option, the external system calls a "get token" endpoint with a username and email, plus its application ID and key, and MI hands back a short-lived API token. The integration then uses that token to make normal REST API calls on behalf of that user. This shows up as a Tokens sub-list, so admins can see issued tokens and when they'll expire.
- [Save]
1.2.2. Configure Additional Features
- Allow Cross Domain Access: Activate this checkbox if the external tool's front-end code runs on a different domain than Metric Insights and needs to call MI's API directly from the browser. When activated, it opens a Cross Domain Origin section where you can whitelist the domains allowed to access it.
- Application Can Manage Users: When this checkbox is enabled, the External Application's tokens can be used to create, view, update, and delete Metric Insights Users through the REST API.
- Click [+ New Redirect URI] and add
https://<mi-host>/mcp/auth/callback, changing<mi-host>to your MI instance name. - [Save] the changes.
Copy Values
- Click [Show credentials].
- Copy Application ID (Client ID) and Application Key (Client Secret) values.
You'll need to collect a Token if you selected the "Token" option in the Auth Type field.
Access Admin > System > API Toolkit
- Item: Select "get_token" option.
- ID: Type anything to open the Request text field.
- Select JSON request.
- Insert Application ID and Application Key in the corresponding fields and type your Username into the
userfield. - [Run Request]
- Collect the Token value from the Raw response field.
2. Configure AI Application
After configuring External Application and collecting all the necessary information, proceed with configuring the AI application your organization uses. For detailed guides, check the following articles: