What problem does it solve? Editing .docx, .xlsx, and .pptx files programmatically usually requires heavy libraries or an Office installation. This Skill lets an AI agent create, read, proofread, and modify Office documents through a single dependency-free CLI, with structured JSON output and a live browser preview. ## Core Features & Use Cases - Layered editing model: L1 read/inspect (view, get, query, validate), L2 DOM operations (add, set, move, remove, batch), and L3 raw XML for anything the DOM layer cannot express. - Live preview and selection: watch serves an auto-refreshing HTML preview where users click shapes or paragraphs, and the agent reads the selection back to apply edits. - Specialized sub-skills: Loadable rule sets for pitch decks, academic papers, financial models, dashboards, and Morph-animated presentations. - Use Case: Ask the agent to build a Q4 sales deck: it creates the .pptx, adds slides and styled shapes, runs view issues to catch formatting problems, and validates the file before delivery. ## Quick Start Ask the agent to create a PowerPoint file named slides.pptx with a title slide and a text shape, then verify it with officecli validate.