PPTX Toolkit — Create, Edit, and Extract PowerPoint Presentations

The problem

Programmatically generating PowerPoint files often results in generic, broken layouts with overlapping text and "AI-standard" design tropes. Handling .pptx files requires a mix of XML manipulation, template parsing, and visual verification that most scripts fail to coordinate.

What it does

• Extracts text and structured data from existing presentations using markitdown.
• Creates new decks from scratch or templates using pptxgenjs and custom editing workflows.
• Automates visual QA by converting slides to images and performing collision detection for text and shapes.
• Applies professionally curated color palettes and typography pairings to avoid default styling.
• Manages complex slide operations like unpacking raw Office XML, thumbnail generation, and PDF conversion for multi-stage processing.

Frameworks & tools

Python (markitdown, Pillow), Node.js (pptxgenjs), LibreOffice, Poppler (pdftoppm), and custom XML unpacking scripts.

Why this beats prompting it yourself

Standard LLM prompting generates code that creates ugly, overlapping slides with leftover placeholder text. This skill enforces a "fix-and-verify" loop that uses subagents to check for visual bugs, margins, and contrast before finalizing the file.

Use cases

• Automating weekly executive reporting decks from raw data sources.
• Parsing incoming pitch decks to extract key stats and summaries for CRM systems.
• Converting technical documentation into structured presentation slides with code syntax highlighting.
• Applying corporate branding and layout motifs to unformatted draft presentations.

Known limitations

Requires LibreOffice and Poppler installed for the visual QA and PDF conversion features to function.