Software product / application
MarkItDown
Convert documents and other files into structured Markdown for language-model inputs and text-analysis pipelines, using a command-line utility or Python API.
About this listing
MarkItDown is Microsoft's Python file-conversion utility for preparing text for language models and analysis pipelines. It handles formats including PDF, Word, PowerPoint, Excel, HTML and common text-based files, aiming to retain document structure such as headings, lists, tables and links.
Use the command-line interface to write Markdown to a file or standard output, or call it from Python as part of an ingestion pipeline. Format-specific optional dependencies let developers enable only the converters they need. The output prioritizes downstream text processing rather than high-fidelity reproduction of the original document.
The code is free under the MIT licence. Local compute and separately selected hosted services are not included: optional image captioning, speech transcription and Azure processing may require external services and incur their own charges. The utility operates with the invoking process's access privileges, so applications handling untrusted inputs should restrict accessible resources and use the narrowest suitable conversion method.