Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.appliedaifoundation.org/llms.txt

Use this file to discover all available pages before exploring further.

​
What it does

The PDF to Markdown skill turns PDFs into searchable, structured markdown using AI vision. It preserves what other extractors lose: table structure, diagrams, form fields, document layout. Once converted, the markdown plus a generated keyword index makes the document instantly searchable across the document library. The skill processes pages in parallel for speed and embeds rendered page images in the markdown so any visual reference in the source remains visible in the converted output.

​
When to use it

  • “Make this searchable”
  • “Extract these PDFs”
  • “Index this folder”
  • “Convert PDF to markdown”
  • “Process these documents”

​
What it preserves

  • Tables — column structure and cell alignment, not flattened text
  • Diagrams — embedded as page images so engineers can still read them
  • Forms — field labels and values kept paired
  • Layout — heading hierarchy, lists, captions
  • Page numbers — every line traces back to its source page

​
How it works

  1. Renders each PDF page as an image.
  2. Calls a vision model with a layout-aware prompt to produce structured markdown for each page.
  3. Runs pages in parallel (configurable workers and rate limit) to keep wall-clock time low.
  4. Stitches per-page markdown into a single document with embedded page images.
  5. Builds a keyword index alongside the markdown for fast search later.

​
What it produces

  • One markdown file per PDF with embedded page images
  • A keyword index for the converted set (drives search-indexed-documents)
  • Metadata mapping markdown back to source PDF and page

​
Modes

  • Single PDF — convert one document
  • Folder, recursive — convert every PDF in a tree, building a unified index
  • Index only — rebuild the search index over an already-converted folder

​
Why vision-based

Text-layer PDFs lose layout when extracted with traditional tools — and many maritime documents are scans with no text layer at all. Running vision over the rendered page captures structure regardless of how the PDF was produced.
  • pdf-vision-extractor — diagram and table extraction from individual pages
  • search-indexed-documents — search across every converted document
  • download-flag-circulars — sister skill that downloads circulars this skill then converts
  • download-makers-circulars — same, for manufacturer circulars