Slide Export

/slide-export — Export HTML Slides to PDF

Export any HTML presentation to a properly sized PDF using decktape. Works with Reveal.js, generic HTML slides, and most other frameworks. Serves slides over a local HTTP server so fonts, images, and external assets load correctly.

Usage

/slide-export [path/to/slides.html]

If no path is passed, ask for the HTML file path before proceeding.

Instructions

Step 1: Pre-flight

Check that decktape is installed:

which decktape

If not found, tell the user and stop:

"decktape is required. Install it with:

• Mac: brew install decktape or npm install -g decktape
• Other: npm install -g decktape Then re-run /slide-export."

Step 2: Locate the file

Use the argument as the HTML file path. If no argument was given, ask: "What's the path to the HTML file?"

Resolve to an absolute path. Confirm it exists:

ls "[absolute_path]"

If it doesn't exist, stop and tell the user.

Step 3: Detect slide dimensions

Read the HTML file. Look for explicit dimensions in this order:

1. Reveal.js config — search for width: and height: inside a Reveal.initialize({...}) block. Extract both integer values.
2. CSS on slide container — look for explicit width and height pixel values on .slides, .slide-deck, or section.present.
3. Viewport meta tag — check <meta name="viewport"> for width/height hints.

If found, use those values as WIDTHxHEIGHT.

If not found, ask: "What size are your slides? Common options: 1920x1080 (16:9 HD), 1280x720 (16:9), 960x700 (Reveal.js default), or enter a custom WIDTHxHEIGHT."

Default if the user skips: 1920x1080.

Step 4: Start a local HTTP server

Serving over HTTP ensures fonts, images, and web-hosted assets render correctly.

Get the directory containing the HTML file. Start Python's built-in server on port 8989:

python3 -m http.server 8989 --directory "[html_directory]" &
echo $!

Capture the PID. Sleep 1 second to let it start.

If port 8989 is in use, try 8990, then 8991.

Step 5: Run decktape

Output path: same directory as the HTML, same filename with .pdf extension.

decktape automatic \
  -s [WIDTH]x[HEIGHT] \
  --load-pause 500 \
  -p 100 \
  http://localhost:[PORT]/[filename] \
  "[output_path]"

Flag rationale:

• automatic — auto-detects the slide framework (Reveal.js, Remark, etc.)
• --load-pause 500 — waits 500ms after page load before capturing (lets fonts and images settle)
• -p 100 — 100ms pause between slides (fast but not choppy)

If decktape fails, show the raw error. If it looks like a framework detection issue, suggest the user retry with decktape reveal instead of automatic.

Step 6: Kill the server

kill [SERVER_PID] 2>/dev/null

Step 7: Confirm

Tell the user the PDF path and open it:

open "[output_path]"

Notes

• Animated slides: If transitions don't fully render, suggest increasing the pause: re-run with -p 500 or higher.
• Custom fonts: If fonts look wrong in the PDF, the slides likely rely on Google Fonts or another CDN. The HTTP server handles this correctly; file:// would not.
• Reveal.js specifically: If automatic fails on a Reveal.js deck, use decktape reveal instead.
• Output location: By default the PDF lands next to the HTML file. If the user wants it somewhere else, ask before running.
• Large decks: decktape processes one slide at a time — large decks (100+ slides) may take a minute.