wtf is document-to-ppt-skill?

Document to PPT Skill is a Codex skill that turns source documents, such as Feishu Wiki pages, Word files, PDFs, Markdown, or plain text, into a presentation deck that looks like it was actually designed rather than just split into slides page by page. It works from formats like Feishu Wiki or Docx for internal training and project reports, Word or PDF for management reports and client proposals, and Markdown or plain text for technical talks and course scripts. It can also take an existing reference PPT or webpage deck and match its structure, pacing, text density, and visual style. The workflow reads the source document's outline and body text, and if a reference deck is given, it studies that deck's style as well. It sorts and prioritizes the content, then produces a page level outline. The skill waits for the user to confirm or adjust this outline before moving forward, and only after confirmation does it generate a Markdown script and the final PPTX file. The final output includes the Markdown script with page titles, visible text, speaker notes, and visual notes, the PPTX file itself matching the reference style, and a quality check covering rendering, overflow, and page number consistency. The core principles listed in the README are to outline first and never generate a final deck without user confirmation, to keep one key point per slide by breaking long paragraphs into definitions, comparisons, processes, cases, or tables, to prioritize matching a reference deck's structure and style when one is given, to avoid overusing visuals and only add charts or diagrams when they genuinely help understanding, and to check the whole deck before delivery for overflow, overlap, broken line breaks, and page numbering. If no reference style is provided, the skill falls back to a low density, white background, grid based, course style look. The skill relies on other capabilities depending on the input type, using a Feishu document capability for Wiki or Docx sources, general document or PDF reading for Word and PDF sources, and a presentations capability for both reading reference PPTX decks and generating the final PPTX. The repository is released under the MIT license.