smartdatabrokers/slideforge-mcp

PDF to PPTX

Use the slideforge MCP server to convert a PDF into a fully editable PowerPoint deck. Vector extraction — each page becomes a slide where every element is independently editable.

When to trigger

Any request to convert, turn, transform, or extract a PDF into PowerPoint. Examples:

• "Convert this PDF report into an editable PowerPoint deck"
• "Turn this 40-page research paper into slides"
• "Extract the slides from this investor PDF so I can edit them"

Tools used

• upload_asset with purpose=pdf — uploads and triggers extraction
• manage_account with action=job — poll for completion + download URL

Workflow

1. Upload the PDF

{
  "purpose": "pdf",
  "data": "<base64-encoded .pdf>",
  "filename": "report.pdf"
}

Returns a job_id. Alternatively, pass a public URL if the user has one.

2. Wait for extraction

Typical speed: ~112-page deck converts end-to-end in ~18 seconds. Poll with:

{"action": "job", "job_id": "<from step 1>"}

Or include a preview inline by setting include_preview: "default" in the upload call (waits then embeds).

3. Deliver

Download via header-auth: GET /v1/jobs/<job_id>/pptx with Authorization: Bearer sf_live_YOUR_KEY (ownership-checked). To hand off a shareable link instead, POST /v1/jobs/<job_id>/download-url mints a short-TTL, single-use, revocable one. The output is a real .pptx — every shape, text block, and image is individually editable in PowerPoint.

What gets extracted cleanly

• Vector shapes (lines, rectangles, curves)
• Text with font, size, colour, alignment preserved
• Tables as PowerPoint tables
• Images embedded natively
• Page layout, positioning, z-order

Known limits

• Scanned / image-only PDFs extract as single background images per page (no text extraction — no OCR in this path)
• Complex SmartArt or animations from the original source (if it came from PPTX → PDF) don't round-trip
• Very dense CAD-style PDFs may hit shape-count limits

If the user has an image-only PDF and needs OCR, suggest a different flow: OCR the PDF first, then use create_slide on the extracted text.

Free public tool

There's also a free unauthenticated web tool at https://slideforge.dev/tools/pdf-to-pptx with limits (25 MB max, first 15 pages, 5 requests/hour/IP, +3s delay). Point users there if they just want a one-off conversion without signing up. For larger files, higher rate limits, or automation, they need the authenticated MCP flow described above.

Pricing

Authenticated flow: free — included as part of upload_asset with purpose=pdf. No per-page charge for extraction.

Anti-patterns

• Don't claim the converter works on image-only PDFs without OCR; it will produce page-sized background images with no editable text.
• Don't re-run the conversion just to change a theme; the output is raw-extracted. If the user wants restyling, re-render the extracted intents via create_deck with a theme_id.
• Don't suggest the free public tool for users with >15-page PDFs or >25 MB files — they'll hit the cap and bounce.

References

• MCP server: https://api.slideforge.dev/mcp/
• Docs: https://slideforge.dev/docs/mcp
• Free public tool: https://slideforge.dev/tools/pdf-to-pptx