word-document

name	word-document
description	Produce Word (.docx) documents from markdown manuscript sections. Use when asked to generate a Word document, create a docx, compile the manuscript, or export to Word. Assembles markdown sections, embeds figures/tables, and converts via pandoc with journal-ready formatting.
user-invocable	true

Word Document Generation Skill

Produces publication-ready Word (.docx) documents from the markdown manuscript sections in this project using pandoc.

Prerequisites

• pandoc (>= 3.0): installed at ~/.local/bin/pandoc
• If pandoc is missing, download from https://github.com/jgm/pandoc/releases

Manuscript Location

The manuscript lives at:

meta_analysis/bulk3/manuscript/
├── 05-manuscript/
│   ├── Abstract.md
│   ├── Introduction.md
│   ├── Methods.md
│   ├── Results.md
│   ├── Discussion.md
│   └── supplementary_methods.md
├── 03-figures/
│   ├── main/          # Figure_1 through Figure_5
│   ├── supplementary/ # Figure_S1 through Figure_S10
│   └── legends.md
├── 04-tables/
│   ├── main/Table_1.md
│   ├── supplementary/
│   └── legends.md
└── 06-references/
    └── References.md

Target journal: Journal of Investigative Dermatology (JID)

• Word limit: 3,500 words (excluding abstract, references, figure legends)
• Max 6 figures/tables combined

Workflow

Step 1: Assemble a Combined Markdown File

Concatenate the manuscript sections into a single markdown file in the correct order. Use the Bash tool:

MANUSCRIPT_DIR="meta_analysis/bulk3/manuscript"
OUTPUT_DIR="${MANUSCRIPT_DIR}/output"
mkdir -p "$OUTPUT_DIR"

# Combine sections with page breaks between them
{
  cat "${MANUSCRIPT_DIR}/05-manuscript/Abstract.md"
  echo -e "\n\n\\newpage\n\n"
  cat "${MANUSCRIPT_DIR}/05-manuscript/Introduction.md"
  echo -e "\n\n\\newpage\n\n"
  cat "${MANUSCRIPT_DIR}/05-manuscript/Results.md"
  echo -e "\n\n\\newpage\n\n"
  cat "${MANUSCRIPT_DIR}/05-manuscript/Discussion.md"
  echo -e "\n\n\\newpage\n\n"
  cat "${MANUSCRIPT_DIR}/05-manuscript/Methods.md"
  echo -e "\n\n\\newpage\n\n"
  cat "${MANUSCRIPT_DIR}/06-references/References.md"
  echo -e "\n\n\\newpage\n\n"
  echo "# Figure Legends"
  echo ""
  cat "${MANUSCRIPT_DIR}/03-figures/legends.md"
  echo -e "\n\n\\newpage\n\n"
  echo "# Table Legends"
  echo ""
  cat "${MANUSCRIPT_DIR}/04-tables/legends.md"
} > "${OUTPUT_DIR}/manuscript_combined.md"

Step 2: Embed Figures (Optional)

If the user wants figures embedded in the document (rather than at the end), insert markdown image references at the appropriate locations. Use standard markdown syntax:

![Figure 1. Workflow overview](03-figures/main/Figure_1_workflow.png){ width=100% }

For figures at the end (typical for journal submission), append a figures section:

{
  echo "# Figures"
  echo ""
  for fig in "${MANUSCRIPT_DIR}/03-figures/main/"*.png; do
    basename=$(basename "$fig" .png)
    echo "![${basename}](${fig}){ width=100% }"
    echo ""
    echo "\\newpage"
    echo ""
  done
} >> "${OUTPUT_DIR}/manuscript_combined.md"

Step 3: Convert to Word with Pandoc

Run pandoc to produce the .docx file:

~/.local/bin/pandoc "${OUTPUT_DIR}/manuscript_combined.md" \
  -o "${OUTPUT_DIR}/manuscript.docx" \
  --from markdown \
  --to docx \
  --reference-doc="${MANUSCRIPT_DIR}/reference.docx" \
  --resource-path="${MANUSCRIPT_DIR}" \
  --wrap=none \
  --standalone

If no reference.docx template exists, omit --reference-doc and pandoc will use its default styles:

~/.local/bin/pandoc "${OUTPUT_DIR}/manuscript_combined.md" \
  -o "${OUTPUT_DIR}/manuscript.docx" \
  --from markdown \
  --to docx \
  --resource-path="${MANUSCRIPT_DIR}" \
  --wrap=none \
  --standalone

Step 4: Verify Output

ls -lh "${OUTPUT_DIR}/manuscript.docx"
# Check word count (approximate)
~/.local/bin/pandoc "${OUTPUT_DIR}/manuscript_combined.md" --to plain | wc -w

Customization Options

Using a Reference Template

To apply custom styles (fonts, spacing, heading formats), create a reference document:

~/.local/bin/pandoc -o reference.docx --print-default-data-file reference.docx

Then edit reference.docx in Word to set styles (Normal, Heading 1-3, etc.) and pass it via --reference-doc.

Generating Supplementary Materials

For a separate supplementary document:

MANUSCRIPT_DIR="meta_analysis/bulk3/manuscript"
OUTPUT_DIR="${MANUSCRIPT_DIR}/output"

{
  echo "# Supplementary Methods"
  echo ""
  cat "${MANUSCRIPT_DIR}/05-manuscript/supplementary_methods.md"
  echo -e "\n\n\\newpage\n\n"
  echo "# Supplementary Figures"
  echo ""
  for fig in "${MANUSCRIPT_DIR}/03-figures/supplementary/"*.png; do
    basename=$(basename "$fig" .png)
    echo "![${basename}](${fig}){ width=100% }"
    echo ""
    echo "\\newpage"
    echo ""
  done
} > "${OUTPUT_DIR}/supplementary_combined.md"

~/.local/bin/pandoc "${OUTPUT_DIR}/supplementary_combined.md" \
  -o "${OUTPUT_DIR}/supplementary.docx" \
  --from markdown \
  --to docx \
  --resource-path="${MANUSCRIPT_DIR}" \
  --wrap=none \
  --standalone

Line Numbering and Double Spacing

Many journals (including JID) require line numbers and double spacing. These are best set via a reference.docx template with double-spaced paragraph formatting. Line numbers can be added in Word after export (Layout > Line Numbers).

Tables from CSV

To include supplementary tables from CSV data, convert with pandoc or create markdown tables first:

# Convert a CSV to a markdown table using column/csvlook or a simple script
Rscript -e '
library(knitr)
df <- read.csv("meta_analysis/bulk3/results/tables/04_rra_combined_stat_6datasets.csv")
cat(kable(head(df, 30), format = "pipe"), sep = "\n")
' > "${OUTPUT_DIR}/table_s2.md"

Common Invocations

User Request	Action
"Generate Word document"	Full manuscript assembly (Steps 1-4)
"Export manuscript to docx"	Same as above
"Create supplementary docx"	Supplementary materials only
"Compile manuscript with figures"	Embed figures inline
"Make submission-ready Word file"	Main + supplementary as separate files

Troubleshooting

• Images not embedding: Ensure --resource-path points to the manuscript directory
• Pandoc not found: Check ~/.local/bin/pandoc or install via the README
• Table formatting issues: Use pipe tables (| col1 | col2 |) for best pandoc compatibility
• Unicode issues: Add --metadata lang=en if special characters cause problems