yfedoseev/pdf_oxide ? reverse-engineered prompt

Build me a super fast PDF toolkit that works in Rust and Python, with a simple command line app too.

I want it to open a PDF, extract the text, images, words, lines, tables, and form fields, and also convert pages into clean markdown. It should be able to grab just part of a page if I give it a rectangle, and it should support basic PDF editing things like merging files, bookmarks, links, annotations, and form data. I’d also like an easy API where I can loop through pages, read page sizes, and pull out content lazily without loading everything at once.

Please make it feel polished and reliable, with examples I can run right away, plus a small MCP server so AI tools can use it from Claude or Cursor. If you need to check the current docs for any language binding details, look them up online.