markitdown

markitdown is a usage guide for Microsoft's MarkItDown, a Python utility that converts a wide range of file formats to Markdown optimized for LLM and text-analysis pipelines. It addresses the common preprocessing problem of getting heterogeneous documents into clean, structure-preserving Markdown that language models can consume, keeping headings, lists, tables, and links intact rather than producing purely human-readable output.

The skill documents the supported formats (PDF, Word, PowerPoint, Excel, images with EXIF and OCR, audio transcription, HTML, YouTube URLs, Wikipedia, RSS/Atom, CSV, JSON, XML, Jupyter notebooks, ZIP archives, EPub, and Outlook MSG) and covers installation with optional dependency extras, command-line usage, and the Python API. It shows practical patterns such as piping input, batch-converting a directory, preparing documents as LLM context, and optional enrichment via an OpenAI LLM client for image descriptions or Azure Document Intelligence for complex PDFs, including the relevant environment variables like OPENAI_API_KEY and AZURE_DOCUMENT_INTELLIGENCE_KEY.

It targets developers and data engineers building document-processing or LLM-preprocessing pipelines, and anyone who needs to extract text from mixed document types into Markdown. All shell examples are read-only conversions (including a curl that downloads a document and pipes it through the converter), and API keys are documented as standard environment variables, so the skill is a benign tooling reference.