origin-pptx

Generates fully editable PowerPoint decks via PptxGenJS native layouts combined with AI-generated illustrations.

Updated Dec 9, 2025
One-click install
npx skills add https://github.com/Aki2022/skills --skill origin-pptx-aki2022
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: origin-pptx
Source: https://github.com/Aki2022/skills/tree/main/origin-pptx
Command: npx skills add https://github.com/Aki2022/skills --skill origin-pptx-aki2022

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pptxgenjs, python-pptx, and includes scripts (resource) and references (resource) components.

What problem does it solve? Creating polished, on-brand PowerPoint presentations with AI typically produces either flat text dumps or uneditable image-based slides. This Skill builds PPTX files where all text, charts, tables, and shapes remain fully editable in PowerPoint, while AI-generated illustrations and Material Symbols icons add visual richness. ## Core Features & Use Cases - Hybrid generation pipeline: A staged workflow (outline → design mockup approval → native build → render verification → human image review → final PPTX) that keeps text and numbers as the single source of truth in outline.md while mockups define composition. - Native charts and tables: Real data is rendered as Excel-editable OOXML charts via addChart and tables via addTable, never baked into images, with a strict gray-gradation brand system (tokens.json) and BIZ UDPGothic typography. - Edit mode for existing decks: Detects whether a deck's source of truth is the build script or a human-edited PPTX, then applies differential rebuilds or surgical python-pptx edits without destroying manual changes. - Use Case: Ask for a 15-slide Japanese business proposal deck; the Skill co-creates the outline, generates design mockups for approval, builds each slide natively with icons and charts, verifies rendering against mockups, and delivers a template-injected final PPTX. ## Quick Start Create a PowerPoint presentation about our Q3 sales results using the pptx skill, with an outline I can review before you build the slides.

Frequently Asked Questions about origin-pptx

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a PowerPoint presentation with AI that stays editable?

Use a hybrid approach where text, shapes, charts, and tables are built natively with PptxGenJS while only illustrations come from image generation. This Skill implements that pipeline so every element remains editable in PowerPoint after generation.

How to add Excel-editable charts to PPTX with pptxgenjs?

Use slide.addChart with real data to produce native OOXML charts that PowerPoint can edit via its embedded workbook. Decorations beyond chart features, like end-of-line labels or callouts, are layered as native text and shapes on top rather than baked into an image.

Can AI image generation draw accurate data charts for slides?

No. Image generation models fabricate numbers, axes, and ratios, so real data charts must be rendered natively. Image generation is only permitted for decorative or illustrative chart-like visuals explicitly marked as not real data.

How do I edit an existing PPTX without losing manual changes?

First compare the artifact against the last build output to determine whether the script or the human-edited file is canonical. If a human edited the PPTX directly, apply surgical python-pptx edits on a copy instead of rebuilding, which would erase all manual work.

Why does PowerPoint ask to repair my generated PPTX file?

pptxgenjs 4.0.1 emits schema violations in line, scatter, and radar chart XML and negative extents for upward or leftward lines. Running the included sanitize_pptx.py script after every build fixes these issues, which LibreOffice rendering alone cannot detect.

What fonts and environment are required for Japanese PPTX generation?

The deck uses BIZ UDPGothic uniformly, standard on Windows 10+ and installable from Google Fonts on macOS. You also need pptxgenjs via npm, LibreOffice with poppler for render verification, and a logged-in Codex CLI for image generation.