magnusthovsland

document-productivity-workflows

"Umbrella workflow for document-centric productivity: OCR/extraction, PDF edits, PowerPoint decks, and office file manipulation."

magnusthovsland 0 Actualitzat fa 3 mesos

Recursos

1
GitHub

Instal·la

npx skillscat add magnusthovsland/hermes-agent-mac/document-productivity-workflows

Instal·la des del registre de SkillsCat.

SKILL.md

Document Productivity Workflows

Use this skill for document-centric work: extracting text from PDFs/scans, editing PDFs, creating or modifying PowerPoint decks, and manipulating Office files.

Choose the path

  • Need text from a PDF/scan → use OCR/document extraction first (pymupdf for text PDFs, marker/OCR paths for scanned or layout-heavy documents).
  • Need a small PDF text/title correction → use nano-pdf-style natural-language edits, then verify the resulting PDF.
  • Need a presentation → create/read/edit .pptx, slides, notes, and templates with PowerPoint scripts; verify the deck structure and, when possible, render thumbnails.
  • Need generic Office package work → inspect/unpack the OOXML package carefully; preserve relationships and content types.

Operating rules

  1. Keep originals unchanged; write outputs with explicit new filenames unless the user requests overwrite.
  2. Verify real artifacts: file exists, page/slide count is plausible, extracted text is non-empty, or modified PDF/deck opens structurally.
  3. For scanned documents, report OCR uncertainty and any pages that failed.
  4. For decks, prefer programmatic generation/editing over manual XML surgery unless the task requires low-level OOXML changes.
  5. Include final paths and the exact command/script used unless the user explicitly asks for only the downloadable artifact.
  6. Direct-download developer handoffs: when the user asks for a direct, downloadable solution and says to omit other relevant information, create one self-contained artifact containing only the prescriptive implementation (requirements, code, and acceptance criteria). Deliver the file without repeating the analysis, alternatives, caveats, or a long chat summary.

Labeled playbooks

OCR and extraction

Use the preserved ocr-and-documents scripts under references/packages/ocr-and-documents/scripts/ for PyMuPDF or marker extraction. Choose based on whether the PDF has embedded text or scanned pages.

PDF text edits

Use nano-pdf recipes for targeted typo/title edits. Validate output and avoid claiming perfect layout preservation without inspection.

PowerPoint

Use the preserved PowerPoint package under references/packages/powerpoint/, including scripts and Office schemas, for deck creation and OOXML-safe manipulation.

Preserved source packages

Full prior skill packages are preserved under references/packages/: nano-pdf, ocr-and-documents, and powerpoint.