Deploy Logic Apps as MCP Servers on Azure Container Apps for AI Workflows

Microsoft’s Logic Apps ecosystem now supports the Model Context Protocol (MCP) server standard, enabling developers to expose workflows as programmable tools for AI agents. By containerizing Logic Apps Standard workflows as MCP servers and deploying them on Azure Container Apps, teams can integrate arithmetic, data processing, and business logic directly into agent-driven workflows. This approach eliminates the need for custom orchestration while leveraging Azure’s scalable infrastructure.

Transforming Logic Apps Workflows into MCP Servers

Logic Apps Standard workflows can function as MCP servers when configured with the correct runtime settings. A Docker container bundles multiple HTTP-triggered workflows—such as addition, subtraction, multiplication, division, modulus, exponentiation, and square root—each acting as a discrete MCP tool. These workflows are automatically exposed as discoverable tools once the MCP server endpoint is enabled in the host.json configuration file with a minimal setup:

{
  "workflow": {
    "McpServerEndpoints": {
      "enable": true,
      "authentication": {
        "type": "anonymous"
      }
    }
  }
}

With this configuration, the container exposes an MCP-compatible endpoint at /api/mcp, making all bundled workflows instantly accessible as tools. No additional code or external services are required; the container itself serves as the MCP server, deployable to Azure Container Apps for cloud-based accessibility.

Deploying the MCP Server on Azure Container Apps

Azure Container Apps provides a managed environment for hosting containerized Logic Apps MCP servers. The deployment process mirrors standard container orchestration but leverages Azure’s built-in scaling and networking capabilities. Once deployed, the server endpoint becomes reachable via a public URL, such as:

`

This URL serves as the entry point for MCP clients, including AI agents in Logic Apps. The server’s tool discovery mechanism ensures all workflows—add, subtract, multiply, divide, and others—are automatically listed as available tools, ready for integration into agent-driven workflows.

Integrating MCP Servers with AI Agents in Logic Apps

A secondary Logic App, designated as an agent loop, can consume the MCP server tools to perform dynamic calculations. Configured with an Azure OpenAI agent, this Logic App receives an arithmetic expression via an HTTP trigger and uses the built-in MCP client connector to interact with the server. The agent autonomously breaks down the expression according to standard mathematical precedence (BODMAS/PEMDAS) and invokes the appropriate MCP tools in sequence.

For example, processing the expression (2 + 3) * 4^2 / 2 proceeds as follows:

• Step 1: Invoke the add tool to compute (2 + 3) → 5
• Step 2: Invoke the pow tool to compute 4^2 → 16
• Step 3: Invoke the mul tool to compute 5 * 16 → 80
• Step 4: Invoke the div tool to compute 80 / 2 → 40

The agent handles the orchestration logic, calling each tool as needed without requiring manual workflow configuration in the consuming Logic App. Each tool invocation triggers a real workflow run on the MCP server container, returning results in real time.

Under the Hood: How MCP Session Management Works

The MCP client connector in Logic Apps handles session initialization, tool discovery, and JSON-RPC protocol framing automatically. When an agent connects to the server, the client sends an initialize request to establish a session, followed by a tools/list request to retrieve the catalog of available tools. Subsequent tools/call requests execute specific workflows, with responses returned in JSON-RPC format.

For developers building custom clients or seeking deeper insight into MCP session mechanics, Azure Logic Apps documentation provides manual walkthroughs of the initialization and tool invocation process. Understanding these underlying mechanisms can help optimize agent workflows and troubleshoot integration challenges.

Scaling MCP Workflows for Enterprise Use Cases

Containerizing Logic Apps as MCP servers unlocks new possibilities for enterprise automation. Teams can deploy specialized workflows—such as data transformation, API aggregation, or business rule evaluation—as discoverable tools for AI agents. This approach reduces development time by abstracting complex logic into reusable MCP components while maintaining scalability through Azure Container Apps.

As AI-driven automation continues to evolve, integrating Logic Apps with MCP servers positions enterprises to leverage agentic workflows without sacrificing control over business logic or data governance. Future updates to Logic Apps and MCP standards may further streamline this integration, enabling even more sophisticated agent interactions.