Diagram Generator

Problem

Engineers often need to turn verbal requirements, review notes, or documentation drafts into maintainable diagrams, but hand-writing Draw.io, Mermaid, or Excalidraw files tends to cause rework around node IDs, container nesting, edge connections, and coordinate layout. This skill targets that workflow: it first constrains natural language input into a structured JSON spec, then generates chart files in three formats, reducing the low-level description cost between “draw an architecture diagram” and a usable file.

How it works

• Intent capture: before creating a chart, confirm diagram type, output format, layout direction, use case, and filename; when editing an existing file, read the target file and apply changes.
• Playbook routing: network topology, system architecture, flowchart, swimlane, UML, and whiteboard sketches each use their own strategy document, avoiding mixed format rules.
• Spec construction: organize nodes, containers, and edges in an elements array, with unique IDs, top-level edges, source / target pointing to defined nodes, and colors in #RRGGBB.
• Quality gate: check format, hierarchy, container bounds, edge connections, and geometry fields before generation; complex Draw.io and Excalidraw charts require explicit geometry.
• Service call: submit diagram_spec through mcp__mcp-diagram-generator__generate_diagram, then the service validates the schema, creates directories, and writes to diagrams/{format}/ or a specified path.

Boundaries and caveats

This skill does not handle bitmaps, SVG vector editing, PDF layout, PPT creation, 3D modeling, or animated visualization. Output quality depends on prompt detail and JSON spec completeness; missing connector tools, unwritable output directories, or mismatched file extensions can trigger tool, schema, or file-write errors.