AI Foundry knowledge base

SaravananSubramanian-5233 0 Reputation points
2026-03-06T12:58:14.6866667+00:00

Hi there,

I am trying to create a knowledge base as my DevOps organisation in foundry project, I have seen in MS training it will have an option on Model Context Protocol (MCP) in the drop down, which I am is not in the list. I dont have github option as well.

Ref to this blog Microsoft Mechanics video

https://www.youtube.com/watch?v=bHL1jbWjJUc
User's image

Regards

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


2 answers

Sort by: Most helpful
  1. Anshika Varshney 15,535 Reputation points Microsoft External Staff Moderator
    2026-03-09T17:15:27.88+00:00

    Hi SaravananSubramanian-5233,

    You are seeing is expected behavior in Azure AI Foundry, and it is mostly a misunderstanding of where Model Context Protocol and GitHub fit in the flow.

    In Foundry, a knowledge base and Model Context Protocol are used at different stages and they do different jobs.

    When you create a knowledge base in Foundry IQ, the system only shows supported knowledge sources such as Azure AI Search, Blob Storage, SharePoint, Web, or OneLake. Model Context Protocol does not appear here because it is not a knowledge source. This is why you do not see MCP or GitHub in the knowledge base dropdown. This is by design and not a configuration issue.

    Model Context Protocol is used at the agent level, not at the knowledge base creation level. After your knowledge base is created, it is exposed to agents through an MCP tool called knowledge base retrieve. This tool is automatically available when you connect a Foundry IQ knowledge base to a Foundry Agent Service agent. That is where MCP comes into play.

    A simple way to think about it is that the knowledge base stores and retrieves data, while MCP is how an agent talks to that knowledge base and other tools.

    If you are following demos or videos, they usually show MCP being configured when setting up an agent, not when creating the knowledge base itself. That is why the UI you see looks different from the training video.

    To make this work end to end, you can follow this approach.

    1. Create your knowledge base in Foundry IQ using a supported source like Azure AI Search. Azure AI Search is required for Foundry IQ knowledge bases and handles indexing and retrieval. [learn.microsoft.com]
    2. Create or open an agent in Foundry Agent Service and connect your knowledge base to the agent. The agent will automatically use the MCP knowledge base retrieve tool to query the data. This is where MCP is used, even though it does not show up as a dropdown in the knowledge base screen.
    3. Third, if you want to use GitHub or other external systems, those are added as MCP tools or external MCP servers on the agent, not as knowledge sources in the knowledge base UI. That is why GitHub does not appear during knowledge base creation.

    For reference, this official document explains the supported design and flow.

    Microsoft Foundry documentation and agent concepts [learn.microsoft.com]

    In short: nothing is missing from your setup. MCP and GitHub are not supposed to appear when creating the knowledge base. They are configured later at the agent level, which matches the current documented behavior of Azure AI Foundry.

    Please let me know if the issue persists after these checks. If you have any remaining questions or need additional details, I’ll be glad to provide further clarification or guidance. If the above steps resolve your issue, kindly confirm.

    Thankyou!

    Was this answer helpful?

    0 comments No comments

  2. AI answer

    2026-03-06T12:58:33.4866667+00:00

    Model Context Protocol (MCP) is used in Microsoft Foundry at the agent/tool layer, not as a knowledge source type inside a Foundry IQ knowledge base.

    From the available documentation, a typical pattern is:

    1. Create a knowledge base backed by Azure AI Search and other supported knowledge sources (Azure AI Search index, SharePoint, OneLake, etc.).
    2. Expose that knowledge base to agents via MCP as a tool. Azure AI Search knowledge bases expose the knowledge_base_retrieve MCP tool for agent integration.
    3. In Foundry Agent Service, configure an agent that uses this MCP tool to query the knowledge base.

    In other words, the MCP option appears when configuring tools for an agent (or when connecting to hosted MCP servers such as Microsoft Learn, Microsoft Sentinel, or Cosmos DB MCP toolkits), not when defining the knowledge base itself. The knowledge base UI will list only data/knowledge sources like Azure AI Search, SharePoint, Blob Storage, Web, and OneLake.

    To use MCP with a knowledge base in a DevOps organization scenario, the supported flow is:

    1. Create the knowledge base in Foundry IQ using supported knowledge sources.
    2. Connect that knowledge base to a Foundry agent via the MCP tool (knowledge_base_retrieve) as described in the "Connect a Foundry IQ knowledge base to Foundry Agent Service" guidance.
    3. Optionally, integrate other MCP servers (for example, Microsoft Learn MCP Server, Microsoft Sentinel MCP tools, or Azure Cosmos DB MCP toolkit) as additional tools on the agent, not as knowledge sources.

    If the dropdown in the knowledge base creation screen does not show MCP or GitHub, that is expected based on the current documented capabilities; those are configured as MCP tools or external MCP servers at the agent level instead of as knowledge sources.


    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

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.