ganoro/df1ef9f0-3138-4b76-8be9-a0e40bc4ccef/claude-plugin/skills/output-generation/docx-generator

DOCX Generator

Generate Word documents from structured report data using python-docx.

Before generating, read the design system files:

• /shared/plugins/nestbox-df1ef9f0/skills/design-system/references/tokens.md — colors, typography, spacing
• /shared/plugins/nestbox-df1ef9f0/skills/design-system/references/components.md — component patterns
• /shared/plugins/nestbox-df1ef9f0/skills/design-system/references/language.md — terminology, disclaimers
• /shared/plugins/nestbox-df1ef9f0/brand/assets/brand-overrides.json — firm name, confidentiality notice

Copy the logo into the workspace: cp /shared/plugins/nestbox-df1ef9f0/brand/assets/logo.png ./logo.png 2>/dev/null || true

Process

Step 1 — Gather Content

Ensure all report content is finalized in _research/ markdown files before generating.

Step 2 — Write Python Script

Build a Python script that uses python-docx. Pull all color values, font sizes, and spacing from the design system tokens — do not hardcode.

import os, shutil
from docx import Document
from docx.shared import Inches, Pt, RGBColor
from docx.enum.text import WD_ALIGN_PARAGRAPH
from docx.enum.table import WD_TABLE_ALIGNMENT

# ── Design Tokens (from design-system/references/tokens.md) ──
PRIMARY     = RGBColor(0x1B, 0x3A, 0x5C)
ACCENT      = RGBColor(0x2E, 0x75, 0xB6)
TEXT        = RGBColor(0x2D, 0x2D, 0x2D)
TEXT_SEC    = RGBColor(0x66, 0x66, 0x66)
TEXT_INV    = RGBColor(0xFF, 0xFF, 0xFF)
SURFACE     = RGBColor(0xF7, 0xF8, 0xFA)
POSITIVE    = RGBColor(0x1A, 0x7A, 0x3A)
WARNING     = RGBColor(0xC6, 0x77, 0x00)
NEGATIVE    = RGBColor(0xC4, 0x26, 0x1D)
CRITICAL    = RGBColor(0x8B, 0x00, 0x00)

# ── Logo ──
LOGO_SRC = "/shared/plugins/nestbox-df1ef9f0/brand/assets/logo.png"
LOGO_LOCAL = "logo.png"
if os.path.exists(LOGO_SRC):
    shutil.copy2(LOGO_SRC, LOGO_LOCAL)

def add_logo_to_doc(doc, width=Inches(2.0)):
    """Insert the firm logo at the current position in the document."""
    if os.path.exists(LOGO_LOCAL):
        doc.add_picture(LOGO_LOCAL, width=width)

Step 3 — Execute and Verify

pip install python-docx 2>/dev/null
python generate_docx.py
ls -la output/report.docx

Save output to output/report.docx.

Document Structure

Page Setup

• Letter size (8.5 × 11 in)
• Margins: Top 1.0 in, Bottom 1.0 in, Left 1.25 in, Right 1.0 in
• Font: Calibri throughout

Required Sections

1. Title page — call add_logo_to_doc(doc) first, then add title, subtitle, date, and "CONFIDENTIAL"
2. Table of contents — placeholder with update-field instruction
3. Report body — sections with H1/H2/H3 hierarchy per tokens
4. Disclaimers — standard disclaimer text from references/language.md

Headers and Footers

• Header: Fund name (left), "CONFIDENTIAL" (right), 8pt, text-secondary
• Footer: "Page X of Y" (center), report date (right), 8pt
• Thin rule separating header/footer from content

Table Formatting

Follow the Data Table component pattern:

• Header row: primary fill, white Bold 10pt text
• Alternating rows: white / surface
• Right-align numeric columns
• Total rows: Bold with border-strong top border
• Source footnote below: text-secondary, 8pt

Number Formatting

Per tokens.md:

• Currency: $X,XXX or $X.XM
• Percentages: X.X%
• Multiples: X.XXx
• Negative: (X,XXX) in negative color

Important Notes

• Always include confidentiality notice in header
• Always include as-of date on title page
• Include source citations for all data tables
• Label all figures as audited or unaudited
• Include standard disclaimers at end (see references/language.md)
• Apply confidence labels per language guide (VERIFIED/REPORTED/ESTIMATED/ASSUMED)