MCP Server that interacts with Azure AI Foundry (experimental)

@azure-ai-foundry

About MCP Server that interacts with Azure AI Foundry (experimental)

A MCP Server for Azure AI Foundry: it's now moved to cloud, check the new Foundry MCP Server

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "mcp_foundry_server": {
      "command": "uvx",
      "args": [
        "--prerelease=allow",
        "--from",
        "git+https://github.com/azure-ai-foundry/mcp-foundry.git",
        "run-azure-ai-foundry-mcp"
      ]
    }
  }
}

Tools

No tools detected

Fetch the live tool list by running this server in a temporary sandbox using the button above.

Overview

What is MCP Server that interacts with Azure AI Foundry (experimental)?

This is an experimental Model Context Protocol server that provides tools for interacting with Azure AI Foundry, including model discovery, deployment, knowledge indexing, evaluation, and fine-tuning. It has been superseded by the cloud-hosted Foundry MCP Server (preview) and this repository will not be updated further.

How to use MCP Server that interacts with Azure AI Foundry (experimental)?

Install uv, set environment variables in a .env file, and configure the server in VS Code via .vscode/mcp.json using the command uvx --prerelease=allow --from git+https://github.com/azure-ai-foundry/mcp-foundry.git run-azure-ai-foundry-mcp. Then start the server and use GitHub Copilot Chat in Agent mode. Alternatively, use the provided GitHub template or the one-click VS Code install link.

Key features of MCP Server that interacts with Azure AI Foundry (experimental)

List, get details, and compare models from Azure AI Foundry catalog.
Create Azure AI Services accounts and deploy models.
Manage AI Search indexes, documents, indexers, and data sources.
Run text and agent evaluations with configurable evaluators.
List, query, and evaluate Azure AI Agents.
Monitor and manage fine-tuning jobs, metrics, and files.

Use cases of MCP Server that interacts with Azure AI Foundry (experimental)

Discover and select suitable models for a project from the model catalog.
Deploy OpenAI models on Azure AI Services and check quotas.
Build and test prototypes using models with GitHub tokens.
Create and query knowledge indexes in Azure AI Search.
Evaluate agent responses with built-in or custom evaluators.

FAQ from MCP Server that interacts with Azure AI Foundry (experimental)

Is this server still maintained?

No—this experimental server has moved to the cloud as Foundry MCP Server (preview). This GitHub repo contains outdated information and will not be updated.

What runtime does the server require?

You need uv installed (see Installing uv). The server runs via uvx with the --prerelease=allow flag.

Where are environment variables set?

They are defined in a .env file in the workspace root, which is referenced in the MCP server configuration (--envFile ${workspaceFolder}/.env).

Which transports or authentication does the server use?

The server uses stdio transport. Authentication is configured via environment variables (e.g., Azure credentials). The new cloud version uses Microsoft Entra ID with On-Behalf-Of flow.

What are the main tool categories?

The server provides tools for Models (explore, build, deploy), Knowledge (index, document, query), Evaluation (text and agent evaluators), and Finetuning (job management, metrics, Swagger actions).

Frequently asked questions

Is this server still maintained?

No—this experimental server has moved to the cloud as Foundry MCP Server (preview). This GitHub repo contains outdated information and will not be updated.

What runtime does the server require?

You need `uv` installed (see [Installing uv](https://docs.astral.sh/uv/getting-started/installation/)). The server runs via `uvx` with the `--prerelease=allow` flag.

Where are environment variables set?

They are defined in a `.env` file in the workspace root, which is referenced in the MCP server configuration (`--envFile ${workspaceFolder}/.env`).

Which transports or authentication does the server use?

The server uses stdio transport. Authentication is configured via environment variables (e.g., Azure credentials). The new cloud version uses Microsoft Entra ID with On-Behalf-Of flow.

What are the main tool categories?

The server provides tools for Models (explore, build, deploy), Knowledge (index, document, query), Evaluation (text and agent evaluators), and Finetuning (job management, metrics, Swagger actions).

Comments

Basic information

More Cloud & Infrastructure MCP servers

🐳 docker-mcp

QuantGeekDev

A docker MCP Server (modelcontextprotocol)

490

Run Model Context Protocol (MCP) servers with AWS Lambda

awslabs

Run existing Model Context Protocol (MCP) stdio-based servers in AWS Lambda functions

373

Kilo IoT Platform

Kilo IoT

Your AI can finally reach the physical world. Kilo is an AI-first IoT platform for commercial and industrial operations, and this connector hands a live deployment to the AI client you already use. Not an export. Not a

MCP Gateway

mcp-ecosystem

🧩 MCP Gateway - A lightweight gateway service that instantly transforms existing MCP Servers and APIs into MCP servers with zero code changes. Features Docker deployment and management UI, requiring no infrastructure modifications.

2.2K

AWS MCP

RafalWilinski

Talk with your AWS using Claude. Model Context Protocol (MCP) server for AWS. Better Amazon Q alternative.

295