domains / documents / document-workflow-routing / document

document

Skill card

combined

The three copies are not reproductions of one procedure. The first is a dispatcher for office file formats (PDF/DOCX/XLSX/PPTX) that routes to other skills. The second is a specific writer for software change documentation (PR, changelog, release notes, postmortems) generated from git history. The third is a catalog of academic/research document processing tools (GROBID, Zotero, PDF parsing). They overlap only in the broad label 'document' but have disjoint steps, tools, and outputs.

Repositories

Other copies

Combined skill

---
name: document
allowed-tools: Bash, Read, Grep, Glob, Write, Edit, Agent, AskUserQuestion
description: "Unified document workflow dispatcher. Routes to office format skills, software change documentation, or academic research tools based on user intent."
---

# Document Workflow Dispatcher

Use this skill when the task involves document creation, processing, or routing. Determine the specific domain and dispatch accordingly.

## 1. Determine Document Domain

Ask the user or infer from context which of the following applies:
1. **Office Formats**: PDF, DOCX, XLSX, PPTX (formatting, conversion, layout).
2. **Software Change Docs**: PR descriptions, changelogs, release notes, postmortems (from git history).
3. **Academic/Research**: Citation parsing, reference extraction, academic PDF processing, Zotero management.

## 2. Dispatch Rules

### A. Office Formats (PDF/DOCX/XLSX/PPTX)
- **PDF**: Use `pdf` skill for extraction, rendering, or layout review.
- **Word/DOCX**: Use `docx` skill for formatting and tracked changes.
- **Excel/XLSX**: Use `xlsx` skill for spreadsheets and pivot tables.
- **Slides/PPTX**: Use `scientific-slides` or `pptx-posters` for decks and posters.
- **Safety**: Do not guess formats. If unclear, ask: "What is the target file type?"
- **Sequencing**: For multiple outputs, sequence as: source-of-truth → derived exports.

### B. Software Change Documentation
- **Types**: `pr`, `changelog`, `release-note`, `postmortem`.
- **Source**: Git commits, diffs, and incident facts.
- **Output Style**: Plain, simple language. No dashes or hyphens in prose (except in code/paths).
- **Process**:
  1. Determine type (ask if unclear).
  2. Gather source material (`git log`, `git diff`, `git tag`).
  3. Write document to the appropriate location (`CHANGELOG.md`, `docs/releases/`, etc.).
  4. For PRs, check `gh` availability and existing PRs.

### C. Academic/Research Tools
- **Citation Parsing**: Use `anystyle-api` or `grobid-pdf-parsing`.
- **PDF Extraction**: Use `pdf-extraction-guide` or `paper-parse-guide`.
- **Reference Management**: Use `zotero-*` skills for Zotero plugins and themes.
- **Long Documents**: Use `large-document-reader` for chapter-by-chapter analysis.

## 3. Output Expectations

- **Office**: Preserve formatting if a template is provided. State output paths and verification steps.
- **Software**: Trace every sentence to a commit or diff. Do not invent timelines or causes.
- **Academic**: Ensure structured extraction of metadata and references. Verify citation accuracy.

## Conflicts

- **jsmastery-pro/skills** vs **foryourhealth111-pixel/Vibe-Skills**: The former writes prose from git history; the latter routes to format-specific skills. They do not share steps.
- **brycewang-stanford/Auto-Empirical-Research-Skills** vs **jsmastery-pro/skills**: The former focuses on academic tools (GROBID, Zotero); the latter on software change docs. They do not share steps.
- **foryourhealth111-pixel/Vibe-Skills** vs **brycewang-stanford/Auto-Empirical-Research-Skills**: The former handles office formats; the latter handles academic PDFs. They may overlap in PDF handling but with different tools and goals.