This is for when you need actual rendered images, not text diagrams. It uses Nano Banana Pro via the Gemini CLI to generate PNGs and SVGs of ERDs, UI mockups, wireframes, and architecture illustrations with a Salesforce architect aesthetic. The workflow is sensible: draft at 1K to check composition, iterate with /edit commands to save tokens, then render final at 4K once you're happy. It gates on prerequisites and defaults to an interview mode that asks clarifying questions before generating, unless you say "quick" or "just generate." Delegates text Mermaid diagrams to a different skill and hands off to LWC or Apex skills once the visual is approved. Honest take: the draft-first approach will save you from expensive 4K regenerations when the layout is wrong.
npx skills add https://github.com/forcedotcom/afv-library --skill generating-visual-diagrams