| name | markitdown |
| description | Convert office / PDF / HTML / image formats to clean Markdown. Prefer this before summarizing document content or extracting structured requirements from attachments. |
| source | https://github.com/microsoft/markitdown |
| integration | external-library |
| status | descriptor-only |
markitdown
Descriptor-only registration (P9 batch). Parallel to the existing skills/document/ (Docling) skill, markitdown is Microsoft's lighter-weight equivalent — good fallback when Docling is unavailable.
Integration plan (deferred)
- Add an optional Python subprocess adapter
octocode_markitdown behind a feature flag.
- Route to
markitdown when the file is .docx/.pptx/.xlsx and Docling is not installed.
- Keep Docling as the default for PDFs (better layout preservation).
Contract
- Input: file path.
- Output: UTF-8 Markdown string +
{source_path, source_mime, generated_at} metadata.