burowski/markitdown ? reverse-engineered prompt

Build me a Python tool that converts lots of common files into clean Markdown so I can feed them into LLMs and text analysis workflows.

I want it to handle PDFs, Word docs, PowerPoints, Excel files, images with OCR and metadata, audio with transcription, HTML, CSV, JSON, XML, ZIP files by reading what’s inside, YouTube links, and ebooks too. The output should keep the important structure, like headings, lists, tables, links, and other readable formatting, instead of dumping plain text.

Please make it usable from the command line and also as a simple library I can call from Python. It should support writing to a file or printing to stdout, and it would be great if it can work with optional extras for different file types instead of forcing everything at once. If useful, include a plugin system and make sure the security around file and network access is handled carefully.