1. User initiates diagram creation, replication, or editing via command or natural language.
2. Skill interprets the user's intent and gathers necessary input (text description, image, edit instructions).
3. Skill leverages the design system to select appropriate themes and semantic shapes.
4. Skill generates or modifies the Draw.io diagram XML.
5. MCP server renders the diagram in a real-time browser preview.
6. User reviews and provides feedback or further edit requests.
7. Skill updates the diagram based on feedback.
8. User exports the final diagram.