mingrammer/diagrams

Diagrams is a Python library that lets you draw cloud architecture diagrams entirely in code, no drag-and-drop design tools required. Instead of opening a diagramming application and manually placing boxes and arrows, you write a short Python script that describes your system's components and how they connect, and the library renders a visual diagram for you. The problem it solves is that architecture diagrams created in visual tools are static images. They are hard to update, easy to let drift out of sync with the actual system, and cannot be reviewed or tracked the same way code can. By expressing the diagram as Python code, you can store it in version control alongside your codebase, review it in pull requests, diff changes, and regenerate the image automatically whenever the architecture changes. Under the hood, Diagrams uses Graphviz, a widely used open-source graph layout engine, to do the actual rendering. You write Python objects representing cloud services (an AWS EC2 instance, a GCP Kubernetes cluster, an Azure load balancer, and so on) and group or connect them with Python syntax like nested context managers and the right-arrow operator. The library ships with icon sets for all major cloud providers including AWS, Azure, GCP, Kubernetes, Alibaba Cloud, Oracle Cloud, DigitalOcean, and many more, as well as generic and on-premises components. You would use Diagrams when designing a new system and want to quickly sketch and share architecture proposals, or when you need to document an existing cloud infrastructure in a format that lives next to your code rather than in a separate slide deck. It is also useful for auto-generating architecture diagrams in CI/CD pipelines or documentation sites, Apache Airflow, for example, uses it to generate diagrams in their official documentation. The stack is Python 3.9 or later with Graphviz installed as a system dependency.