This guide is for running slides-grab with Codex and shared Vercel-installed skills.
npm install slides-grab
npx playwright install chromiumVerify:
slides-grab --helpFrom the local npm install:
npx skills add ./node_modules/slides-grab -g -a codex --yes --copyThen restart Codex so the installed skills are loaded.
If you want to work on slides-grab itself instead of consuming the package:
git clone https://github.com/vkehfdl1/slides-grab.git && cd slides-grab
npm ci
npx playwright install chromium
npx skills add . -g -a codex --yes --copyInstalled skill names:
slides-grab-planslides-grab-designslides-grab-exportslides-grab
Run one deck per workspace folder:
Prerequisite: create or generate decks/my-deck/ with slide-*.html first. These commands do not work against an empty clone.
slides-grab edit --slides-dir decks/my-deck
slides-grab build-viewer --slides-dir decks/my-deck
slides-grab validate --slides-dir decks/my-deck
slides-grab image --slides-dir decks/my-deck --prompt "Editorial hero image for slide 03"
slides-grab pdf --slides-dir decks/my-deck --output decks/my-deck.pdf
slides-grab pdf --slides-dir decks/my-deck --mode print --output decks/my-deck-searchable.pdf
slides-grab convert --slides-dir decks/my-deck --output decks/my-deck.pptx
slides-grab figma --slides-dir decks/my-deck --output decks/my-deck-figma.pptxslides-grab pdf defaults to --mode capture for browser-faithful rendering. Switch to --mode print when you need searchable/selectable text in the exported PDF.
When a slide needs bespoke generated imagery, use slides-grab image --prompt "..." --slides-dir <path> with GOOGLE_API_KEY (or GEMINI_API_KEY) set. If no key is available, ask for one or fall back to web search + local download into assets/.
Copy-paste into Codex:
Read docs/installation/codex.md first and follow it exactly. Install slides-grab with npm, install the shared skills from ./node_modules/slides-grab using Vercel Agent Skills, use slides-grab-plan/design/export (or slides-grab), keep each deck in decks/<deck-name>, and run validate before convert/pdf.