About this skill

Problem

When natural-language descriptions need to become sequence, class, activity, use case, component, deployment, state, Gantt, or mind-map diagrams, PlantUML output can be hard to reuse because styling, syntax, and rendering setup vary. This skill connects requirement parsing, source generation, and image rendering to produce maintainable *.puml files and reviewable outputs.

How It Works

Diagram selection: infer sequence, class, activity, use case, component, deployment, state, Gantt, or mind-map diagrams from signals such as request/response, inherits/fields, approval flow, permissions, or deployment nodes; ask for clarification when ambiguous.
PlantUML generation: emit @startuml and @enduml, include a descriptive title, avoid decorative styling, and enforce a black-and-white, thin-line, shadow-free UML visual baseline.
Rendering: use the bundled script to render through the PlantUML public server, Docker, or a local plantuml.jar, producing svg, png, pdf, or text; show inline svg where possible and report saved files otherwise.
Iteration: update the .puml source based on feedback and re-render to keep code and image consistent.

Limits and Notes

Good for turning text into structured diagrams and standardizing UML style; not ideal for branded visuals, animation, or pixel-perfect layout.
The monochrome line-art style is the default; request explicit deviations for color themes, decorative corners, or custom icons.
Public-server rendering depends on network access; offline use requires Docker or Java with plantuml.jar as fallbacks.