andrewmoshu/diagram-mcp-server

Model Context Protocol (MCP) server for Multi-Cloud Infrastructure Diagrams
This MCP server seamlessly creates diagrams using the Python diagrams package DSL. Generate professional infrastructure diagrams for any cloud provider (AWS, GCP, Azure), Kubernetes, on-premises, hybrid, and multi-cloud architectures using natural language with Claude Desktop or other MCP clients.
> Note: This is a derivative work based on awslabs/aws-diagram-mcp-server, extended with multi-cloud provider support and enhanced features.
## Prerequisites
1. Install GraphViz with development headers - Required for diagram generation and .drawio export
2. Install uv from Astral (recommended) or use pip
3. (Optional) Install Helm - Required for full Helm chart parsing with parsehelmchart
### Installing GraphViz
macOS (Homebrew):
bash
brew install graphviz
Ubuntu/Debian:
bash
sudo apt-get install graphviz graphviz-dev
Windows (Chocolatey):
powershell
choco install graphviz
Or download from graphviz.org
### Installing Helm (Optional)
Helm is required for full template rendering with the parsehelmchart tool. Without Helm, the parser will use a fallback mode with limited template support.
macOS (Homebrew):
bash
brew install helm
Ubuntu/Debian:
bash
curl https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash
Windows (Chocolatey):
powershell
choco install kubernetes-helm
Or download from helm.sh/docs/intro/install
## MCP Client Configuration
Configure the MCP server in your MCP client. The CFLAGS and LDFLAGS (or INCLUDE and LIB on Windows) environment variables are needed for the initial build of pygraphviz.
Claude Desktop / Cursor on macOS (~/Library/Application Support/Claude/claudedesktopconfig.json or ~/.cursor/mcp.json):
json
"mcpServers": {
"infrastructure-diagrams": {
"command": "uvx",
"args": ["infrastructure-diagram-mcp-server"],
"env": {
"FASTMCPLOGLEVEL": "ERROR",
"CFLAGS": "-I/opt/homebrew/include",
"LDFLAGS": "-L/opt/homebrew/lib"
> Note: /opt/homebrew is the default Homebrew prefix on Apple Silicon Macs. For Intel Macs, use /usr/local instead.
Linux (e.g., ~/.config/claude/claudedesktopconfig.json):
json
"mcpServers": {
"infrastructure-diagrams": {
"command": "uvx",
"args": ["infrastructure-diagram-mcp-server"],
"env": {
"FASTMCPLOGLEVEL": "ERROR"
> Note: On Linux, install graphviz-dev first (sudo apt-get install graphviz graphviz-dev), then no extra env vars are needed.
Windows (e.g., %APPDATA%\Claude\claudedesktopconfig.json):
json
"mcpServers": {
"infrastructure-diagrams": {
"command": "uvx",
"args": ["infrastructure-diagram-mcp-server"],
"env": {
"FASTMCPLOGLEVEL": "ERROR",
"INCLUDE": "C:\\Program Files\\Graphviz\\include",
"LIB": "C:\\Program Files\\Graphviz\\lib"
> Note: Adjust the Graphviz path if you installed it in a different location.
Other MCP Clients (e.g., Kiro - ~/.kiro/settings/mcp.json):
json
"mcpServers": {
"infrastructure-diagrams": {
"command": "uvx",
"args": ["infrastructure-diagram-mcp-server"],
"env": {
"FASTMCPLOGLEVEL": "ERROR",
"CFLAGS": "-I/opt/homebrew/include",
"LDFLAGS": "-L/opt/homebrew/lib"
},
"autoApprove": [],
"disabled": false
### Docker Installation
Build and run with Docker:
bash
docker build -t infrastructure-diagram-mcp-server .