AI Potluck
Product / UX / Agent tools & protocols

MarkItDown

Microsoft

Lightweight Python utility from Microsoft for converting files and office documents - PDF, DOCX, PPTX, XLSX, images (OCR), audio (transcription), HTML, CSV/JSON/XML, ZIP, EPub, even YouTube URLs - into Markdown optimized for LLMs and agents. Offers CLI, Python API, Docker, and a plugin system, with optional Azure Document Intelligence and LLM-vision for image descriptions.

The PyPI license field is null, so the terms come from the repository and pyproject, which is what the openness axis reads. The Azure Document Intelligence and Content Understanding paths are optional install extras rather than a core withheld from the package. Verified 2026-08-13 via GitHub, the repository LICENSE and the pypistats API.

Openness

5 high confidence
5.0
license
MIT(OSI
source
public
core-gated
ungated

Fully MIT; optional Azure/LLM features are pluggable, not gating. Azure Document Intelligence and Azure Content Understanding are optional install extras - the `[az-doc-intel]` and `[az-content-understanding]` groups - sitting beside built-in offline converters, so the conversion the product exists to do runs entirely in the published package and core-gated stays ungated.

  • https://raw.githubusercontent.com/microsoft/markitdown/main/LICENSE recorded 2026-08-13

    Plain MIT License text: "Copyright (c) Microsoft Corporation".

  • https://github.com/microsoft/markitdown recorded 2026-08-13

    Repo page - MIT badge, root tree carrying LICENSE, Dockerfile and packages/. A capability table compares "Built-in converters" (offline, format-specific) against Azure Document Intelligence and Azure Content Understanding, both installed as optional extras `[az-doc-intel]` and `[az-content-understanding]`. No license key, second license or enterprise directory.

Adoption

5 medium confidence
5.0

12,755,994 downloads of markitdown in the trailing 30 days, which bands at >10M, level 5 on the software adoption scale. The count is for the PyPI package, not the unrelated npm package of the same name.

Capability

4 medium confidence
4.0

Broadest format coverage among RAG ingestion tools; lighter on deep layout parsing than Docling.

  • https://github.com/microsoft/markitdown recorded 2026-08-13

    Format list - PDF, Office documents, images (OCR), audio (speech transcription), HTML, text-based formats (CSV, JSON, XML), "ZIP files (iterates over contents)", YouTube URLs, EPubs. Optional extras `[audio-transcription]`, `[youtube-transcription]`, `[az-doc-intel]`, `[az-content-understanding]`; a Plugins section; a Dockerfile in the root tree.

Verified 2026-08-12