Share via

where do i find the endpoint and api key for an AI Foundry AI Agent

Trevor Gardiner 5 Reputation points
2025-07-01T21:56:50.56+00:00

I have created an AI Agent in Azure AI Foundry. I have given a system prompt and uploaded Knowledge documents. I cannot find the Endpoint URL and API key

Foundry Tools
Foundry Tools

Formerly known as Azure AI Services or Azure Cognitive Services is a unified collection of prebuilt AI capabilities within the Microsoft Foundry platform


1 answer

Sort by: Most helpful
  1. Amira Bedhiafi 41,966 Reputation points MVP Volunteer Moderator
    2025-07-02T09:11:27.8666667+00:00

    Hello Trevor !

    Thank you for posting on Microsoft Learn.

    Open the Azure AI Foundry portal, then navigate to your specific project then go to the Overview or Libraries section.

    Look for your project endpoint, it will have the format:

    https://<YourFoundryResource>.services.ai.azure.com/api/projects/<YourProjectName>
    

    This is the base URL you'll use in SDKs or REST calls.

    https://learn.microsoft.com/en-us/azure/ai-foundry/agents/quickstart

    Within the same project in the portal, find the Settings, Keys and endpoints, or sometimes named Connected resources.

    There should be an option like Keys and endpoints or Manage keys, click there to view the API keys for your resource.

    https://www.lamatic.ai/docs/models/providers/azure/azure-ai-foundry

    https://help.specinnovations.com/azure-api-key

    If it's not visible yet:

    • The service stores API keys in an associated Azure Key Vault.
    • You may need Owner, Contributor, or Cognitive Services Contributor permissions to view them.

    https://www.huuhka.net/wrapping-your-head-around-azure-ai-foundry

    https://learn.microsoft.com/en-us/azure/ai-foundry/agents/quickstart

    Once there, you can copy the API key string. You can also regenerate it, or switch to Microsoft Entra ID (managed identity) later for stronger security.

    Was this answer helpful?

    0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.