Generative-AI agents only become useful when they can do things—query systems of record, trigger workflows, or look up specialized knowledge. Until now that meant hand-rolling Azure Functions, managing OpenAPI specs, or writing custom plug-ins for every backend you own. MCP changes the economics: it is an open, JSON-RPC–based protocol—originally proposed by Anthropic—that lets a “server” publish tools (functions) and resources (context) once and have any compliant “client” (your agent runtime) discover and call them automatically. Think “USB-C for AI integrations.” With today’s preview, Foundry Agent Service become first-class MCP clients. Bring any remote MCP server—self-hosted or SaaS—and Azure AI Foundry will import its capabilities in seconds, keep them updated, and route calls through the service’s enterprise envelope. Model Context Protocol (MCP) allows developers, organizations, and service providers to host services and APIs on MCP servers and easily expose and connect tools to MCP-compatible clients, such as the Foundry Agent Service. MCP is an open standard that defines how services provide various functions and context to AI models and agents. With Foundry Agent Service's support of MCP, users can bring an existing MCP server endpoint and add it as a tool to Foundry agents. When connecting to an MCP server, actions and knowledge are automatically added to the agent and updated as functionality evolves. This streamlines the process of building agents and reduces the time required for maintaining them. With MCP support in Foundry Agent Service, it empowers you to: Easily integrate with services and APIs. Whether you want to connect with your internal services or APIs from external providers, MCP provides an easy way to integrate with Foundry Agent Service without writing and managing custom functions. Enhance your AI agent with enterprise features in Foundry Agent Service. With Foundry Agent Service, you can enable enterprise-ready features such as Bring Your Own thread storage.Your browser does not support the video tag.Code SamplesStep 1: Import the needed packagesimport timeimport jsonfrom azure.ai.agents.models import MessageTextContent, ListSortOrderfrom azure.ai.projects import AIProjectClientfrom azure.identity import DefaultAzureCredentialStep 2: create AI Project Client and create an Azure AI Foundry Agentserver_labelis a unique name you provide for the MCP server within the same Foundry Agentserver_urlis the url of the MCP serverrequire_approvalsupports neverproject_client = AIProjectClient( endpoint=PROJECT_ENDPOINT, credential=DefaultAzureCredential())with project_client: agent = project_client.agents.create_agent( model=MODEL_DEPLOYMENT_NAME, name="my-mcp-agent", instructions="You are a helpful assistant. Use the tools provided to answer the user's questions. Be sure to cite your sources.", tools= [ { "type": "mcp","server_label": , "server_url": , "require_approval": "never" } ], tool_resources=None ) print(f"Created agent, agent ID: {agent.id}")Step 3: Create a Thread, Message and RunWithin the Run, you can pass custom headers and use the server_label you provided to map to the specific MCP server. thread = project_client.agents.threads.create() print(f"Created thread, thread ID: {thread.id}") message = project_client.agents.messages.create( thread_id=thread.id, role="user", content="", ) print(f"Created message, message ID: {message.id}") run = project_client.agents.runs.create(thread_id=thread.id, agent_id=agent.id)Step 4: Execute the Run and retrieve MessageYou can use run_step to get more details on tool inputs and tool outputs. Once the run is complete, you can retrieve message back from the Foundry Agent. # Poll the run as long as run status is queued or in progress while run.status in ["queued", "in_progress", "requires_action"]: # Wait for a second time.sleep(1) run = project_client.agents.runs.get(thread_id=thread.id, run_id=run.id) print(f"Run status: {run.status}") if run.status == "failed": print(f"Run error: {run.last_error}") run_steps = project_client.agents.run_steps.list(thread_id=thread.id, run_id=run.id) for step in run_steps: print(f"Run step: {step.id}, status: {step.status}, type: {step.type}") if step.type == "tool_calls": print(f"Tool call details:") for tool_call in step.step_details.tool_calls: print(json.dumps(tool_call.as_dict(), indent=2)) messages = project_client.agents.messages.list(thread_id=thread.id, order=ListSortOrder.ASCENDING) for data_point in messages: last_message_content = data_point.content[-1] if isinstance(last_message_content, MessageTextContent): print(f"{data_point.role}: {last_message_content.text.value}")Step 5: Clean up project_client.agents.delete_agent(agent.id) print(f"Deleted agent, agent ID: {agent.id}")At Microsoft Build 2025, Satya Nadella highlighted an “open-by-design” AI ecosystem and announced that Microsoft is partnering with Anthropic to make the Model Context Protocol (MCP) a first-class standard across Windows 11, GitHub, Copilot Studio, and Azure AI Foundry. Today’s preview support for MCP in Azure AI Foundry Agent Service is the next step in that journey. It brings the same “connect once, integrate anywhere” promise to cloud-hosted agents, letting you take any MCP server and plug it directly into Foundry with zero custom code. Create with Azure AI Foundry Get started with Azure AI Foundry, and jump directly into Visual Studio Code Download the Azure AI Foundry SDK Read documentation to learn more about the feature Take the Azure AI Foundry learn courses Keep the conversation going in GitHub and Discord