What problem does it solve? Japanese-style Excel documents (方眼紙, 帳票, 画面設計書) rely on heavy merged cells and visual layout, so naive spreadsheet-to-Markdown tools flatten them into repeated, unreadable tables. This Skill converts such document-like workbooks into HTML with rowspan/colspan plus semantic Markdown, while recording every limitation in a report. ## Core Features & Use Cases - Workbook classification: Inspect each sheet's title, merge count, and used range to pick a conversion strategy (cover-kv, control-table, merged-doc, table-dense, layout-asset) before converting. - HTML + Markdown export: Emit per-sheet HTML preserving merged-cell geometry and Markdown rebuilt semantically, including detection of コントロール名/コントロールID control tables. - convert-report.json: Always produce a machine-readable report listing per-sheet strategies, export sizes, and explicit limitations (shapes, charts, truncation, stamp heuristics). - Use Case: Given a JP screen design workbook like 画面設計書_FBD13001.xlsx, classify it, convert all sheets, and hand the Markdown to a RAG pipeline while keeping the HTML as layout ground truth. ## Quick Start Ask the agent to run the excel-doc-convert skill to classify and then convert your .xlsx design document into HTML and Markdown with a convert-report.json in the output folder.