Bonus Lab - Add Knowldege capability to Trey Genie using a Microsoft Graph connector

Do these labs if you want to build a Declarative agent where Microsoft 365 provides the AI model and orchestration

This lab builds on the Lab E4. You should be able to continue working in the same folder for labs E2-E6, but solution folders have been provided for your reference. The finished Trey Genie declarative solution for this lab is in the /src/extend-m365-copilot/path-e-bonus-gc-lab/trey-research-labEB-END folder. The Microsoft Graph connector source code is in /src/extend-m365-copilot/path-e-bonus-gc-lab/trey-feedback-connector folder.

In this lab you will learn to:

deploy a Microsoft Graph connector of your own data into Microsoft Graph and have it power various Microsoft 365 experiences
customise the trey genie declarative agent to use the Graph connector as a capability to extend its knowledge
learn how to run and test your app

Prerequisites: Tenant Admin Access

Additonal prerequisites are needed to run this lab. You will need tenant administrator privileges as Microsoft Graph connectors use app-only authentication to access the connector APIs.

Exercise 1 : Deploy Graph Connector

Step 1: Download sample project

In your browser, go to this link
Extract the trey-feedback-connector.zip file

Note

The extracted folder of the sample project is trey-feedback-connector. It has a folder called content which consist of feedback files from various clients for consultants at Trey Research. The files are all created by AI and are for demo purposes only. The aim is to deploy these external files into Microsoft 365 data to be available as knowledge base for our declarative agent Trey Genie.

Step 2: Create external connection

Open the folder trey-feedback-connector in Visual Studio Code
In the Activity Bar of Visual Studio Code, open the Teams Toolkit extension
Create a file .env.local in the env folder of the root folder trey-feedback-connector
Paste below contents in the newly created file

APP_NAME=TreyFeedbackConnectorApp
CONNECTOR_ID=tfcfeedback
CONNECTOR_NAME=Trey Feedback Connector
CONNECTOR_DESCRIPTION=The Trey Feedback Connector seamlessly integrate feedback data from various clients about consultants in Trey Research.
CONNECTOR_BASE_URL=https://localhost:3000/

Select F5, which will then kick off the creation of the Entra ID app registration needed for your connector API to authenicate and load data into Microsoft Graph
In the Terminal window, for the func:host start Task, you will notice below link provided. Using this link you can grant the app-only permission for the Entra ID app

entra id consent link

Copy the link and open in a browser where you are logged in as the tenant admin for the Microsoft 365 tenant.
Grant the required permissions to the app using the Grant admin consent button.

entra id grant

Once granted, the connector creates an external connection, provisions the schema and ingests the sample contents in the content folder to your Microsoft 365 tenant. This takes a while, so keep the project running.
Once all files in the content folder is loaded, the debugger can be stopped.
You can also close this connector project folder.

Step 3: Test the connector data in Microsoft365 app

Now that your data is loaded into Microsoft 365 tenant, let's test if a regular search is picking up the contents in Microsoft365.com.

Go to https://www.microsoft365.com/ and in the search box above, type thanks Avery.

You will see the results as below from the external connection which are basically the clients' feedback for consultant Avery Howard.

Now that your data is part of Microsoft 365 data or Microsoft Graph, let's go ahead and add this connector data as focused knowledge for our declarative agent for Trey Research called Trey Genie.

Exercise 2 : Add Graph Connector to Declarative Agent

In the previous exercise, we established a new external connection to load our data into the Microsoft 365 tenant. Next, we will integrate this connector into our declarative agent to provide focused knowledge on Trey Research consultants.

Step 1: Get the connection id of the Microsoft Graph Connector

In exercise 1, we added the environment variable in the .env.local file which has the configuration values for the Graph connector. The connection id value we gave is tfcfeedback. When Teams Toolkit deploys this connector it will add a suffix of its environment value like local to the connection id. Hence we can infer the connection id is tfcfeedbacklocal. But the most straight forward way to get the Graph connector id is to use Graph Explorer.

Browse to Microsoft Graph Explorer and sign in with your admin account.
Select your user avatar in the upper right corner and select Consent to permissions.
Search for ExternalConnection.Read.All and select Consent for that permission. Follow the prompts to grant consent.
Enter https://graph.microsoft.com/v1.0/external/connections?$select=id,name in the request field and select Run query.
Locate the connector you want and copy its id property.

connection id

Step 2: Update declarative agent manifest

Let's now resume with our declarative agent from Lab 4. If you have it open, then continue or go to the finished lab 4 solution in this folder /src/extend-m365-copilot/path-e-lab04-enhance-api-plugin/trey-research-lab04-END.

Open the lab 4 solution for Trey Genie declarative agent.
Go to appPackage\trey-declarative-agent.json
Add a new item into the capabilities array as below and save

 {
            "name": "GraphConnectors",
            "connections": [
                {
                    "connection_id": "tfcfeedbacklocal"
                }
            ]
}

Now the capability is added, it's time to test.

Exercise 3: Test the plugin in Copilot

Step 1: Start the application

Update the manifest version of your app package in file appPackage\manifest.json from "version": "1.0.0" to "version": "1.0.1". This will make sure the changes are applied.

Start your project to force it to re-deploy the application package by selecting F5. You will be brought into Microsoft Teams. Once you're back in Copilot, open the right flyout 1️⃣ to show your previous chats and agents and select the Trey Genie Local agent 3️⃣.

Running the declarative agent

Step 2: Test knowledge in Copilot

In the immersive experience of Trey Genie, use below prompts and test them

Can you check for any feedback from clients for consultants Trey Research
How did Avery's guidance specifically streamline the product development process?

Working

CONGRATULATIONS!

You have completed lab Graph connector lab, Well done!