Converts entire PDFs to structured Markdown in one shot using IBM's Docling AI, preserving headers, tables, formatting, and images with relative paths. First run is slow (about a second per page) but aggressively caches everything so subsequent loads are instant. The auto-view behavior is smart: when you ask about charts or diagrams, it'll read the extracted images without prompting. Best for when you need the whole document in context rather than grepping specific pages. Uses TableFormer AI for table extraction with 93% accuracy on complex layouts, and cranks image resolution to 4x for clarity. Ships with its own venv to stay out of your project dependencies.
npx skills add https://github.com/aliceisjustplaying/claude-resources-monorepo --skill pdf-to-markdown