appautomaton/document-skills
4 skills3.3k installs428 starsGitHub
Install
npx skills add https://github.com/appautomaton/document-skillsSkills in this repo
1Xlsxxlsx is an agent skill packaged as a Python script that recalculates every formula in an Excel workbook so downstream steps see computed values, not stale cached cells. It uses openpyxl to manipulate files and LibreOffice in headless mode with a StarBasic macro that calls `calculateAll()`, stores, and closes the document. Solo builders adopt it when agents generate financial models, importers, or reporting templates where openpyxl alone leaves formulas unevaluated. The skill auto-creates the LibreOffice macro directory on first run when needed. It suits CLI-style document automation in Build and recurring Operate reporting jobs, not interactive spreadsheet editing in the browser.1.1kinstalls2DocxDocx is an agent skill that teaches solo builders to generate Microsoft Word .docx files using the JavaScript docx library in Node.js or the browser. It belongs on the Build docs shelf when you need contracts, reports, spec exports, or client deliverables that must open cleanly in Word rather than as pasted markdown. The tutorial stresses reading the full guide first because small mistakes—like newline characters inside TextRun—can corrupt documents or break rendering. It walks through setup with npm install, constructing Document sections, and saving via Packer buffers or blobs. Coverage includes text formatting, tables, images, headers and footers, table of contents, footnotes, hyperlinks, and page layout options. Use it when an agent should output polished Office artifacts for validate proposals, build documentation, or grow content workflows. It is a generator-style integration skill grounded in one library, not a general PDF or Google Docs toolchain.1.1kinstalls3PptxPPTX is an agent skill for working with Microsoft PowerPoint files (.pptx) as first-class artifacts in a repo or chat session. Solo and indie builders use it when they need investor decks, launch slides, workshop materials, or stakeholder reviews without context-switching to desktop Office. The skill distinguishes lightweight reading (convert to markdown for text-only analysis) from deep edits that require unpacking the ZIP-based OOXML package and manipulating XML under `ppt/`. That split matters for agents that must preserve speaker notes, slide masters, comments, or non-trivial layout. It fits builders shipping SaaS or content products who still produce slide-based narratives for validate, launch, and grow milestones. Expect shell access to Python/uv tooling and filesystem paths to the unpack script relative to the project root.647installs4PdfThe PDF skill is a document-handling agent capability for solo builders who need reliable PDF form workflows without guessing library APIs. It forces a critical first step: determine whether a PDF has fillable AcroForm fields using the bundled check script, then branch into fillable-field extraction (field_id, page, rect, type) or non-fillable handling per the skill instructions. Operators run uv-run Python utilities from the skill directory to emit field_info.json and complete forms systematically rather than improvising one-off scripts that corrupt layouts or miss checkbox value pairs. The skill fits agents automating onboarding packets, government forms, invoices, or internal PDF templates inside Claude Code, Cursor, or Codex. Expect intermediate complexity: you need uv, Python, and tolerance for PDF coordinate systems. It is a procedural workflow skill, not a generic markdown converter—optimized for correctness on real-world forms where order of operations matters.534installs