drawio-skill
An agent skill that turns natural language, code, Terraform/Kubernetes, SQL, OpenAPI, AsyncAPI, Protobuf and GraphQL sources into editable draw.io diagrams, with incremental sync, multiple views and CI architecture tests.
drawio-skill is an agent skill that turns natural language and real system sources into maintainable .drawio architecture diagrams. It works with Claude Code, Cursor, Copilot, OpenClaw, Codex, Autohand Code, Hermes and any agent that supports the Agent Skills format.
From a text prompt, it plans a diagram's layout, writes the XML, exports it, and checks its own rendered image to fix overlaps, clipped labels and stacked edges automatically, with further rounds of user feedback; it can also convert Mermaid text for 28 diagram types into a laid-out, editable .drawio file, or rebuild an editable diagram from a photo or screenshot of a whiteboard. It offers 11 diagram-type presets, including ERD, UML class and sequence, C4, architecture, flowcharts, SysML, BPMN and network topology diagrams.
From real sources, it visualizes a codebase's import graph for Python, JavaScript/TypeScript, Go and Rust, turns Terraform, Kubernetes and docker-compose configuration, or a live snapshot of deployed infrastructure, into diagrams with official cloud provider icons, and converts SQL DDL, OpenAPI, AsyncAPI, Protocol Buffers, GraphQL SDL and CI pipeline definitions into their corresponding diagrams. It is written in Python under the MIT license.