Markdown File Converter

Problem

When working with source materials, text in PDFs, Office files, HTML, CSV, JSON, and images is often trapped in binary layouts or markup. Copying it by hand can discard headings, tables, lists, and links, while raw files are awkward to feed into search indexes, knowledge bases, or model pipelines. This skill turns those common inputs into Markdown, a text-friendly intermediate format for reading, indexing, and further processing.

How it works

The skill runs uvx markitdown, so the main converter does not require a separate installation. It supports Documents such as PDF, Word (.docx), PowerPoint (.pptx), and Excel (.xlsx / .xls); Web/Data such as HTML, CSV, JSON, and XML; Media such as image EXIF extraction and OCR, plus audio EXIF and transcription; and Other inputs including ZIP contents, YouTube URLs, and EPub files. The output preserves document structure, including headings, tables, lists, and links. The first run caches dependencies, making later runs faster. For complex PDFs or files with poor extraction, the notes suggest using the -d option with Azure Document Intelligence. It fits cases where scattered files should become versionable plain text, but not DRM-protected files, very poor scans, or tasks requiring exact original layout.