document

name	document
title	Document Reading & Authoring
description	Read and author office documents. Use for any task with document reference files (.pdf/.docx/.pptx/.xlsx/.csv) or that asks to produce reports, contracts, decks, spreadsheets, or memos. Extracts text and tables, renders PDF pages to images for visual inspection, and provides thin builders for DOCX/PDF/PPTX/XLSX deliverables.
modalities	["document"]
file_extensions	[".pdf",".docx",".doc",".pptx",".ppt",".xlsx",".xls",".csv",".rtf",".odt"]
keywords	["document","report","memo","contract","brief","letter","deck","slide","presentation","spreadsheet","workbook","table","pdf","docx","pptx","xlsx","invoice","statement","form","page","paragraph","heading"]
requires	["pdfplumber","PyMuPDF","python-docx","python-pptx","openpyxl","pandas"]
version	1.0.0

Document Skill — read & write office files

Use this skill to read reference documents accurately (exact text, tables, and — via page images — layout) and to author polished deliverables.

A typical "document problem" workflow:

1. text = document.read_any(path) — extract text from PDF/DOCX/PPTX/XLSX/CSV.
2. tables = document.pdf_tables(path) — pull tabular data from PDFs.
3. (visual check) imgs = document.pdf_to_images(path) — render pages to PNG so the vision layer / image skill can inspect layout, charts, signatures.
4. Build the deliverable with the appropriate builder.

Toolkit API

from skills import document

document.read_any(path) -> str            # auto-dispatch by extension
document.read_pdf_text(path) -> str       # pdfplumber text
document.pdf_tables(path) -> list[list[list]]   # tables per page
document.pdf_to_images(path, out_dir="pages", dpi=150) -> list[str]  # PyMuPDF
document.read_docx(path) -> str
document.read_pptx(path) -> list[str]     # text per slide
document.read_xlsx(path) -> dict          # {sheet: [rows...]}

document.make_docx(title, paragraphs, out) -> str
document.make_pdf(title, lines, out) -> str
document.make_pptx(title, slides, out) -> str   # slides = [(heading, [bullets])]
document.make_xlsx(sheets, out) -> str          # sheets = {name: [rows...]}

Notes:

• pdf_to_images is the bridge to vision: render pages, then inspect them with the image skill (OCR / colours) or describe them.
• Builders are intentionally thin — for rich formatting use python-docx / python-pptx / reportlab / openpyxl directly (all pre-installed).