excrtx-produce-slides

Provide your Markdown file or brief and the Skill generates three style previews, then renders a self-contained HTML deck on a fixed 1920x1080 stage. The Markdown stays as the canonical source while HTML, PDF, and ZIP are treated as exports.

Run the included extract-pptx.py script to pull text, images, and speaker notes from the .pptx into JSON and an assets folder. The extracted content then goes through style discovery and is regenerated as a styled HTML deck.

Use Marp for routine, code-heavy, or recurring teaching material where maintainability and fast PDF output matter most. Use the HTML renderer when the presentation needs visual impact, narrative direction, or a premium browser deliverable.

Run bash scripts/export-pdf.sh on the HTML file; it serves the deck locally, screenshots each .slide element at 1920x1080 with Playwright, and assembles a PDF. A --compact flag renders at 1280x720 for smaller files, and a Chrome headless fallback exists if Playwright is unavailable.

Private Google Drive upload is the default destination and never targets the Drive root. Public Vercel deploy is an exception that requires an explicit user request, Draft-First approval, and acceptance of the external account dependency.

The export script locates slides by querying elements with the .slide class. Decks generated by this Skill always use that class, so the error means the HTML was created externally or uses a different class name.