MarkItDown

MarkItDown turns a file into Markdown an LLM can read, and it optimises for structure rather than fidelity: headings, lists, tables and links survive, page layout does not. The format list runs through PDF, Word, PowerPoint, Excel, HTML, CSV, JSON, XML, EPub, images with EXIF and OCR, audio with speech transcription, and zip archives it walks through entry by entry. It is a command line tool first, so markitdown report.pdf > report.md is the whole workflow, and piping from stdin works too. Optional dependencies install per format or all at once through the all extra. Microsoft maintains it and warns that conversion runs with the privileges of the calling process, so untrusted input deserves the narrowest convert function you can get away with.