AI generates a real, editable PowerPoint from any document — native shapes & animations, speaker notes voiced as audio narration, and the option to follow your own .pptx template, not slide images
Offline checker for PowerPoint .pptx decks before you present or send them. Slide-by-slide evidence for missing titles, text-heavy slides, tiny fonts, missing alt text, off-slide objects, leftover comments and linked media, plus speaker-notes talk time. No API key; your deck never leaves your machine.
Download package that runs on your own computer inside your AI agent (e.g. Claude Code or Codex — its subscription/API costs are yours). Requires Python 3.9+ and python-pptx; PowerPoint or LibreOffice optional for PDF. No third-party services or extra fees; works offline. Turns a topic, outline or long document into an editable 16:9 PowerPoint: 9 layouts, 6 themes, auto-fit text, CJK-safe fonts, speaker notes. AI-assisted content: check facts.
Transform any PowerPoint into a studio-quality animated video—not a slideshow of screenshots. This skill reconstructs PPT slides into HTML/CSS/SVG for near pixel-perfect fidelity, preserving layouts, text, shapes, arrows, tables, and diagrams. It adds AI voiceover, strict audio-slide synchronization, smart animations, multi-layer QA validation, and automatic correction loops for failed slides. Ideal for sales decks, training, product demos, explainers, and client presentations.
After 10+ years in HR at top companies, I've built more decks than I can count — talent reviews, hiring cases, board updates — and watched which ones actually changed a decision. The winners were never the prettiest; they were the clearest: one point per slide, a title that states the takeaway, a flow that builds to the ask. This Skill turns that standard into your slides — structuring the story, sharpening every headline, and building a clean, professional deck from whatever content you bring.
静态检查授权PPTX的内部资源关系、外部引用和用户声明的空占位，输出逐页JSON证据。需自备CPython 3.11.1；不渲染、不修复文件，不下载外部链接。
按用户明确的页映射，对比两版PPTX的文本、备注、形状几何和媒体SHA，保留未映射页与未知项。需自备CPython 3.11.1；不推断页面重排，不作视觉等同结论。