Incoming request security

[Bao Tran (n1340t)]

Hi teams, our organization wants to adopt Teams Bot at scale but now I am stuck with security team. They require all incoming requests must pass APIM check. The APIM gateway only allows token issuer by our own application registration. However, I believe issuer of Teams token is https://api.botframework.com which is unfamiliar and not recognized by our APIM gateway. Is there any way to overcome this issue? Can I setup service principal with BotFramework and would it help?

[Corina (corinagum)]

Hi Bao Tran (@n1340t), since you're using a cross-tenant system with User managed Identity, I recommend configuring APIM to validate Bot Framework token directly.

 <validate-jwt header-name="Authorization" failed-validation-httpcode="401" failed-validation-error-message="Unauthorized. Access token is missing or  invalid.">
     <openid-config url="https://login.microsoftonline.com/botframework.com/v2.0/.well-known/openid-configuration" />
     <audiences>
         <audience>YOUR_MANAGED_IDENTITY_CLIENT_ID</audience>
     </audiences>
 </validate-jwt>

Make sure to confirm exact aud and iss claims for your bot. For User Managed Identity, I believe the audience is your MI's client ID, but would be good to verify. You may also need to add an <issuers> block if your APIM policy requires explicit issuer validation.

• This doesn't require changing your bot's tenancy model, which is important for cross-tenant configuration
• Compatible with Manged Identity
• OpenID provides Microsoft's signing keys, so tokens are verified like any Entra ID token.
• Your security team can verify this is a Microsoft-operated endpoint

Other options that would be less suitable:

• Single tenant bot: Would make the token issuer your own Entra ID tenant, however you are already using multi-tenant system. Single tenant would require production and dev be on the same tenant
• APIM passthrough (no JWT validation at APIM): Technically this is secure because the bot service validates Bot Framework tokens internally, however it is unlikely to satisfy your security team's requirements.
• IP allowlisting at APIM level: AzureBotService publishes IP ranges, but this is generally considered a less-secure option
• Token exchange proxy: A lightweight proxy between the bot and APIM that validates the token and re-issues one from your production tenant before forwarding through APIM. Adds latency and complexity.

Some links:

• https://learn.microsoft.com/en-us/azure/bot-service/rest-api/bot-framework-rest-connector-authentication?view=azure-bot-service-4.0&tabs=multitenant
• https://learn.microsoft.com/en-us/azure/api-management/validate-jwt-policy

Hope this helps :)