clip-xlsx

Run the clip-xlsx script with the workbook path, optionally passing --source, --quality, and --topic flags. It converts the file via LibreOffice and pandoc, then writes a Markdown clipping with standard frontmatter to raw/clippings/.

PDF conversion flattens tables and rebuilds row-label pairings from horizontal position, which can pair the wrong number with the wrong label. For spreadsheets the pairing is the content, so direct workbook conversion is required.

The conversion requires LibreOffice (soffice) to render sheets as HTML tables and pandoc to convert that HTML to Markdown. A missing LibreOffice install is fatal and must be fixed before clipping.

Yes, duplicates are detected by content hash rather than file path. A renamed or moved workbook whose contents already exist in the vault is recognized instead of being clipped twice under a new slug.

A corrupt or password-protected workbook returns a failed status so batch runs continue, and it is reported for manual handling. A thin result under 100 words records a decline, with no override flag to force it.

Conversion is lossy on formulas, merged cells, charts, and multi-row headers; it captures rendered values, not the underlying model. Any figure should be verified against the original workbook before it appears on a wiki page.