office
the document clerk
route by the requested output and the source format. read a guide only for the operations it actually documents; use the project's existing document conventions and inspect a source artifact before editing it.
formats preservedtracked changes respectedno silent reflow
Use for office document work -- creating/editing/analyzing Word .docx documents (tracked changes, comments, formatting preservation, text extraction), PDF processing (extract text/tables, create, merge, split, form filling), PowerPoint .pptx decks (slides, layouts, speaker notes, template modification), and spreadsheets .xlsx/.csv/.tsv (formulas, pivot tables, charts, data analysis).
methodology
- identify the deliverable format:
.docx, pdf,.pptx, or.xlsx/.csv/.tsv. - read the learned layer, then the matching guide.
- a pipeline across formats (pdf tables into xlsx) chains the guides in data order.
- render and inspect pages or slides when layout matters; verify formulas and cached values when a spreadsheet contains formulas.
- resolve bundled
scripts/andooxml/paths relative to the matching guide's directory, not the working project's root. Check tool availability before relying on LibreOffice, Poppler, or a Python package.
contents
- trainedlearned layer -- read first
- docxword documents -- creation, edits, tracked changes, comments, formatting preservation, extraction
- pdfpdfs -- text/table extraction, creation, merge/split, form filling, batch operations
- pptxpowerpoint decks -- slides, layouts, speaker notes, deck generation and edits
- xlsxspreadsheets (.xlsx/.csv/.tsv) -- formulas, pivot tables, charts, recalculation