We're working on moving some apps from Razor Pages to React, and we'd like to use Microsoft Identity Platform for authentication via MSAL.
Some of these apps are multi-tenant, and we want to be able to validate that a given tenant is set up and authorized when a user logs in. Today, we do that in .NET via the OnTokenValidated event by pulling the email domain or tenant ID from the token claims and comparing it to our app's list of authorized tenants in our databases.
In MSAL, I'm not entirely clear how to accomplish that. The MS docs say, "If you allow multi-tenant authentication, and you do not wish to allow all Microsoft account users to use your application, you must provide your own method of filtering the token issuers to only those tenants who are allowed to login" (in https://github.com/AzureAD/microsoft-authentication-library-for-js/blob/dev/lib/msal-angular/docs/multi-tenant.md#multi-tenant).
Where in the React login or token acquisition process can I inspect the claims in order to perform this filtering? How are others handling these kinds of checks with MSAL on a SPA front-end framework like React?
Related
I am building an application which has a front-end (a SPA built with Vue.js) which interfaces to a couple json-based Web APIs in the back end (hosted in Azure). The Web APIs need to be secured via Azure Active Directory and users must be a member of a security group. Furthermore, the SPA should simply try to force the user to sign into an approved account if they are not signed in as one (i.e. just auto-redirect).
I actually have all this working. The AAD application has Group.Read.All, the user signs in via the SPA and gives consent, and the SPA calls getMemberGroups. Furthermore, the Web APIs can check the SPA-provided access token, and the Web APIs unfortunately must also call getMemberGroups.
And I think that is my concern. The Web APIs keep having to call getMemberGroups to lock it down. If I did the auth on the service, I could potentially only return an access token once membership groups has been verified. But then I lose the easy MSAL sign-in model in the SPA - the Web APIs don't actually provide any front end, the SPA is statically hosted.
As far as I can tell, I cannot get Azure Active Directory to create a token guaranteed to have certain group claims in it. I think this would solve my problem.
Can somebody offer some advice on the best way to design an auth model around a SPA + Web API environment? Or is the method I have taken the only way to do it?
Thanks!
You can include Groups claim in your token as instructed here. You just need to modify the "groupMembershipClaims" field in application manifest:
"groupMembershipClaims": "SecurityGroup"
Then the token will contain the Ids of the groups that the use belongs to like below :
{
"groups": ["1ce9c55a-9826-4e32-871c-a8488144bb32"]
}
You can also leverage Role along with Groups to control access of your application. You can define some applciation roles and assign the roles to the groups. Then the users in the group will have the claim like below:
{
"roles": ["admin"]
}
Then you can implement your authorization logic based on the roles of the user.
Refer to https://joonasw.net/view/using-groups-vs-using-app-roles-in-azure-ad-apps and https://learn.microsoft.com/en-us/azure/active-directory/develop/howto-add-app-roles-in-azure-ad-apps for more details
I'm working on an application that has an angular 6 front end and a .net core 2.0 back-end and am trying to set it up so that my application authenticates users via Azure active directory. The issue is that I want the .net core back-end to do all the authentication programmatically when I have the front end pass an email and password. Everything that I've seen so far online suggests to have the front end redirect to the Microsoft login page but we are using our application as a boiler plate project and want to be able to change the type of authentication easily in the back-end without having to change the angular front-end. Does anyone know how I could go about this?
You can use Resource Owner Password Credentials Grant .The process will like that you collect the user credentials in Angular app and post to .net core back end , then finish the authentication in back-end app with user's credential. But that is not recommended because The ROPC flow requires a high degree of trust and user exposure and you should only use this flow when other, more secure, flows can't be used.
Also :
The Microsoft identity platform endpoint only supports ROPC for Azure AD tenants, not personal accounts. This means that you must use a tenant-specific endpoint (https://login.microsoftonline.com/{TenantId_or_Name}) or the organizations endpoint.
Personal accounts that are invited to an Azure AD tenant can't use ROPC.
Accounts that don't have passwords can't sign in through ROPC. For this scenario, we recommend that you use a different flow for your app instead.
If users need to use multi-factor authentication (MFA) to log in to the application, they will be blocked instead.
Reference : https://learn.microsoft.com/en-us/azure/active-directory/develop/v2-oauth-ropc
Is there a way to prevent users from logging into their own AAD tenants? For example, allow login to contoso.onmicrosoft.com but not fabrikam.onmicrosoft.com.
My customer has a highly-controlled Azure environment where they are running some servers and PaaS/SaaS applications. Users connect to those servers are use a variety of Azure services and some SaaS (primary concern is PowerBI). They’re concerned that a user could login with a non-company account (wahid#hotmail.com) and then upload sensitive data their own PowerBI workspace.
You cannot prevent a user who has already been in the tenant to login to that AAD tenant, except deleting the user from that tenant.
If one account was created/invited to one tenant, the user just can login to that tenant.
For the scenario you described, I suggest you could revoke the Product license for those users. If you don't assign product license to the user, the user cannot use that App or cannot see that app in My Apps panel. For other SaaS/PaaS app(Web App), you may use user assignment to allow specific users to access the App.
For others: Daniel answered this for me. Short answer, this isn't easy, you would need to inspect the response body for the tenant ID (or name) and then block it. This would also lead to a poor user experience but that's all we can do today.
Yes, you have a few options.
Option 1: Single tenant apps
If you have the ability to create apps in the tenant you want to accept users from, you can mark your app as available to only this tenant (the field is called availableToOtherTenants). This will notify the token service to only authorize access from users in the tenant the app was created in.
Option 2: Multi-tenant app w/ token validation
The other option is to mark your app as multi-tenant (same field as above, just set to true) and implement logic in your app to validate the user's tenant from which the token was issued.
In this scenario, you will need to have some kind of web service that can safely validate access tokens (.NET code sample on this). To check the tenant the user's account belongs to, you'll need to validate the iss field. It'll look something like this,
"iss": "https://sts.windows.net/7fe81447-da57-4385-becb-6de57f21477e/"
in which the GUID represents the tenant ID. This allows your web API to have an allow or deny list based off tenant IDs.
More help
Here's an excellent blog post on token validation.
Azure AD Developer Docs
.NET Web API Code Sample
I am building a native iOS application and want to use AADB2C as identity provider where users login, signup, reset their passwords etc.
I cannot figure out a way to let users signup with AADB2C (or regular AAD for that matter) without redirecting them to a (customizable, but still) microsoft website. To be perfectly clear: I want to let customers create user accounts on AAD from a native iOS form without redirecting them to a website, preferably via REST request. (Like here under "Create consumer user accounts": https://learn.microsoft.com/en-us/azure/active-directory-b2c/active-directory-b2c-devquickstarts-graph-dotnet)
Can you create users from an iOS app?
Yes, using the Graph API as per the article you showed. You can only create local accounts at this time though.
However you need to be very careful about how you do it given that currently, the ability to create users requires Directory.ReadWrite.All permission, which also allows all other sorts of operations. You should NOT put the client ID and client secret for an app with these permissions in your iOS app. Rather, you would need to create a backend service that exposes an API for your iOS app to call for user creation.
However, more importantly, what you WON'T be able to do is SIGN IN the users without a redirect (which is what the B2C sign up policy does). In order to do this from your own UI without redirects, you would need Azure AD B2C to support Resource Owner Password Credentials Flow so that you can, after creating the user, use this flow to sign them in and get a token.
Note: You would also need to disable Email Verification so that you can leverage the user account right after user creation. You can set this in the Sign-up policy or Sign-up/Sign-in policy via Page UI customization > Local account sign-up page > Email Address > Require Verification > No
Lastly, as an FYI, there's a feature in the works in Azure AD B2C: Customer Owned Domains, which, paired up with UI customization, would allow you to have sign-up/sign-in pages that you can look like your own and have a URL of your own, with no trace of Microsoft for your end users to see.
I'm currently trying to implement a multi-tenant Azure AD application that will use Microsoft Graph API's to monitor and analyze Office 365 "metadata" for members of the tenant domain. For example, the application might monitor One Drive user space over time. The architecture of the application will include an AngularJS SPA client along with a web application back-end. The idea is that the web application allows for both local registration (e.g. traditional sign up using an email address and password) in addition to Azure AD authentication. In the case of local registration, the user might be able to associate an Azure AD tenancy with the local account in the future, for example.
I'm struggling to understand how various authentication mechanisms should work. For example, I think that there should be two levels of authentication in the case of Azure AD: one authentication for the users of the client SPA, and another authentication used by the back-end for making continuous calls to the Microsoft API's, requesting refresh tokens, etc.
How might this architecture be implemented using the various Azure AD authentication scenarios Microsoft has already provided examples for?
If my initial inclination that I will have two applications registered with Azure AD (for example, the SPA registered as a native application, say, and the web application registered by itself), how will users allow access to both of them, and what would this workflow look like? In addition, what would the flow of user requests look like? The SPA would make a request to the back-end using its Azure AD token, but what will the back-end do to receive its authentication token and make calls to the Microsoft API's?
How might I best incorporate Azure AD authentication along with local registration into my application?
Generally speaking, you can associate your each user to his entity in Azure AD tenant in your backend server / database. As every user in Azure AD has several unique properties in the entity object. You can use the user's email or objectId as mentioned at Claims in Azure AD Security Tokens as the external column in your user table.
When your user authenticate your site via ADAL.JS, you can grab the access token in your backend server via the Authentication header. You can use the access token to request for the resources protected by Azure AD. And the access token is a JWT token, which you can decode directly to get the user basic claims as we mentioned before. You can retrieve the claim which you stored in your user table and match the special user registered in your server for requesting the resource protected by your self.