Markdown to PDF Converter

Convert Markdown documents to professionally styled PDFs using pandoc + weasyprint with customizable CSS templates.

Input

The user provides a Markdown file path (or multiple paths / glob): $ARGUMENTS

If no path is provided, ask the user which .md file to convert.

Dependencies

• pandoc — brew install pandoc
• weasyprint — pip3 install --user weasyprint
• CSS templates — bundled in ${CLAUDE_SKILL_DIR}/templates/

If either tool is not found, print install instructions.

Arguments

Parse $ARGUMENTS for:

• Input file(s) — one or more .md file paths, or a glob pattern like *.md (required)
• --output or -o — custom output path (optional, defaults to ~/Desktop/<filename>.pdf; single file only)
• --title or -t — document title for HTML metadata (optional, defaults to filename)
• --template — template name (optional, defaults to report)

Templates

Available templates (CSS files in ${CLAUDE_SKILL_DIR}/templates/):

Users can also pass a path to any .css file as the template value.

Template resolution order:

1. If value ends in .css and file exists → use directly
2. comms/templates/<name>.css (project-local override)
3. ${CLAUDE_SKILL_DIR}/templates/<name>.css (bundled)

Conversion Steps

Step 1: Locate files and parse arguments

Verify input file(s) exist. If a file is not found, use Glob to search for likely matches.

Step 2: Find CSS template

SKILL_DIR="${CLAUDE_SKILL_DIR}"

# Resolve template
find_css() {
  local name="$1"
  # Direct .css path
  if [[ "$name" == *.css ]] && [[ -f "$name" ]]; then echo "$name"; return; fi
  # Project-local
  if [[ -f "comms/templates/${name}.css" ]]; then echo "comms/templates/${name}.css"; return; fi
  # Skill-bundled
  if [[ -f "${SKILL_DIR}/templates/${name}.css" ]]; then echo "${SKILL_DIR}/templates/${name}.css"; return; fi
}
CSS=$(find_css "${TEMPLATE:-report}")

Step 3: Convert MD to HTML

pandoc "<input.md>" \
  -o "/tmp/md2pdf_temp.html" \
  --standalone \
  --metadata "title=<title>" \
  --css "<css-path>" \
  --embed-resources

Step 4: Convert HTML to PDF

weasyprint \
  "/tmp/md2pdf_temp.html" \
  "<output.pdf>" \
  2>/dev/null

Step 5: Cleanup and report

rm -f /tmp/md2pdf_temp.html

Report:

• Output file path
• File size
• Remind user to open and verify formatting

Batch Mode

If the user provides multiple files or a glob pattern (e.g., *.md):

1. Expand the glob to a list of files
2. Convert each file individually (output to ~/Desktop/<filename>.pdf)
3. Report results in a summary table:

Status  File                                               Size
------  --------------------------------------------------  ----
OK      ~/Desktop/report1.pdf                              245K
OK      ~/Desktop/report2.pdf                              180K
FAIL    broken.md                                          conversion error

Alternatively, use the bundled shell script directly:

bash "${CLAUDE_SKILL_DIR}/scripts/md2pdf.sh" --template report *.md

Shell Script

A standalone scripts/md2pdf.sh is included for use outside Claude Code:

# Single file
bash ~/.claude/skills/pdf/scripts/md2pdf.sh report.md

# With template
bash ~/.claude/skills/pdf/scripts/md2pdf.sh --template minimal report.md

# Batch
bash ~/.claude/skills/pdf/scripts/md2pdf.sh --template report *.md

Troubleshooting

• weasyprint not found: Run pip3 install --user weasyprint (on macOS with managed Python, add --break-system-packages)
• pandoc not found: Run brew install pandoc
• Chinese characters garbled: Ensure CSS has CJK font fallback (PingFang SC, Microsoft YaHei)
• Tables overflowing: CSS should have table { width: 100%; font-size: 10pt; }
• Template not found: Check ls ~/.claude/skills/pdf/templates/ for available templates