Data Vis Coding V2 · hsuan1012 bundle 建立「互動式資料視覺化應用」的完整工作流程 skill。從一份資料（CSV/JSON/GeoJSON）與需求出發， 走完 規格釐清 → 資料處理 → 圖表選型 → 元件實作 → 多層次驗證，產出「可運作、能互動、數值正確、 風格一致」的 React + Tailwind + Chart.js/D3 前端。核心是「內容與樣式分離」兩階段設計，並以實際 執行 + 數值校驗取代臆測，避免資料錯置與視覺幻覺。 Use whenever the user wants to build or scaffold an interactive data-visualization app / dashboard / chart / choropleth or geographic map / scatter or bubble plot / treemap / sunburst / sankey / heatmap — any runnable, data-driven interactive frontend. Trigger even when they just hand over a dataset and a goal without naming a chart type (e.g.「把這份資料做成可以互動的視覺化」「做個 dashboard」 「build a viz app」). Do NOT use for: adjusting or restyling an EXISTING chart (colors, labels) → the dataviz skill; a single static chart image (one matplotlib PNG); pure data cleaning/analysis whose output is numbers or insights, not an app; explaining or teaching viz concepts; or debugging an existing app / learning a framework.
