azure-diagram

This MCP server renders Microsoft Azure diagrams from natural language prompts using the Python diagrams package. It accepts a short description and generates a diagram (PNG) saved in the diagrams/ folder of the repository. The server leverages the diagrams library with Azure icons to produce Azure architecture diagrams, along with support for sequence diagrams, flow charts, class diagrams, and more, depending on the Python code generated from your prompt.

To use it, configure the MCP client to point at the Azure Diagram MCP Server (the Python module azure_diagram_mcp_server.server). Once running, you can trigger Generate Diagram actions via your MCP-capable editor (e.g., VS Code Copilot workflow) by providing a natural-language prompt such as: “Generate an Azure Architecture Diagram with an Azure Function connected to a SQL Database through a Load Balancer.” The server will execute the underlying diagrams code and output a PNG image into the diagrams/ folder for you to review and export.

Prerequisites:

• Python 3.10+ installed on your system
• GraphViz installed (required by the diagrams package for rendering)
• Git to clone the repository

Setup steps:

1. Install Python 3.10+ (if not already installed)

   • Windows: use the official Python installer from python.org or winget
   • macOS/Linux: ensure python3 and pip3 are available

2. Install GraphViz

   • Windows: download and install from graphviz.org
   • macOS: brew install graphviz
   • Linux: apt/yum install graphviz

3. Clone the MCP server repository (or download the ZIP) and open in your editor:

   git clone <repository-url> cd <repo-directory>

4. Install Python dependencies

   • In the project directory, install requirements:

     pip install -r requirements.txt

5. Run the MCP server via the MCP client configuration (see mcp_config in this document) or use the docker option if you prefer containerized execution.

Optional: set up the MCP client configuration as described in the repository README to integrate with VS Code Copilot workflows.

Tips and known considerations:

• Ensure GraphViz executables are in your system PATH so that diagrams can be rendered.
• The server saves generated diagrams to the diagrams/ folder within the repo; keep an eye on disk space if generating many diagrams.
• If Copilot prompts do not produce a diagram, verify permissions and that the MCP server module (azure_diagram_mcp_server.server) is importable from the Python environment used by the MCP client.
• If you prefer to run in Docker, build the image with: docker build -t azure-diagram-mcp-server . and run as shown in the README example.
• The project relies on the diagrams library and Azure icons; ensure network access is allowed for package imports and any download steps during first run.