drawio-skill is a Python tool that generates and maintains draw.io architecture diagrams from text descriptions and real sources while preserving manual layout edits.
The tool addresses the problem of keeping architecture diagrams synchronized with evolving systems without losing manual refinements. It works by converting text and source data into a diagram intermediate representation (IR) with source-kind profiles, then syncing changes incrementally back to draw.io files. This approach allows diagrams to stay current as systems change while respecting any layout adjustments made directly in the diagram editor. The tool supports multiple diagram types including C4 models, UML, ERD, BPMN, SysML, Kubernetes, and Terraform visualizations.
Teams should adopt this tool if they need architecture diagrams that remain maintainable as code and infrastructure evolve. It suits projects where diagrams serve as living documentation rather than static artifacts, particularly those using CI/CD pipelines. The tool's multi-view projection capability lets different stakeholders see relevant subsets of the same architecture, while the Story Mode provides an accessible narrative view. The built-in MCP server integration enables use within agent-based workflows, and the architecture-as-test CI action allows diagrams to be validated as part of automated testing.
The project shows active development with regular commits across multiple areas of the codebase. Work spans core diagram generation logic, CI integration features, and documentation updates. The maintainers respond to issues and incorporate feedback into the tool's capabilities. Development activity indicates ongoing refinement of both the core synchronization mechanism and the surrounding ecosystem of integrations.
