Docx template filling

Skill cxcscmu/SkillLearnBench/skills/b4-skill-creator-gemini-3-flash-preview/offer-letter-generator/docx-template-filling

[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.

Install

npx -y skills add cxcscmu/SkillLearnBench --skill docx-template-filling

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Robustly fill Word (.docx) templates with JSON data. Handles split placeholders, nested tables, headers/footers, and conditional logic like IF/END_IF blocks. Use this skill whenever you need to generate professional documents from templates.

SKILL.md

3.2 KB, as published. Nobody here has run it

docx-template-filling

Overview

This skill provides a robust framework for filling Microsoft Word templates using the python-docx library. It specifically addresses the "Split Placeholder" problem and supports conditional content blocks.

Core Principles

Paragraph-Level Processing: Always modify text at the paragraph level (para.text) rather than the run level (run.text) to avoid issues where Word splits placeholders like {{NAME}} into multiple runs (e.g., {{NA and ME}}).
Formatting Preservation: When replacing text in a paragraph, clear all runs except the first one. Update the first run's text to the new content to preserve the paragraph's initial formatting.
Comprehensive Traversal: Process the main document body, tables (including nested ones), and headers/footers of all sections.

Handling Conditional Sections

Conditional sections are marked with {{IF_CONDITION}}...{{END_IF_CONDITION}}.

If condition is True: Remove the markers but keep the content in between.
If condition is False: Remove the entire block including markers and content.

Implementation Pattern for Conditionals

def process_conditionals(doc, condition_key, is_enabled):
    start_marker = f"{{{{IF_{condition_key}}}}}"
    end_marker = f"{{{{END_IF_{condition_key}}}}}"
    
    # Track paragraphs to delete if condition is False
    paras_to_delete = []
    in_block = False
    
    for para in doc.paragraphs:
        if start_marker in para.text:
            in_block = True
            # Remove marker from this paragraph
            para.text = para.text.replace(start_marker, "")
        
        if in_block and not is_enabled:
            paras_to_delete.append(para)
            
        if end_marker in para.text:
            in_block = False
            # Remove marker from this paragraph
            para.text = para.text.replace(end_marker, "")

Handling Split Placeholders

Use a regex-based replacement strategy at the paragraph level.

import re

def replace_placeholders(para, data):
    text = para.text
    pattern = r"\{\{([A-Z_0-9]+)\}\}"
    matches = re.findall(pattern, text)
    
    if not matches:
        return False
        
    new_text = text
    for key in matches:
        placeholder = f"{{{{{key}}}}}"
        if key in data:
            new_text = new_text.replace(placeholder, str(data[key]))
            
    if new_text != text:
        # Preserve first run formatting
        if para.runs:
            para.runs[0].text = new_text
            for run in para.runs[1:]:
                run.text = ""
        else:
            para.text = new_text
        return True
    return False

Traversal Logic

Ensure all parts of the document are reached:

doc.paragraphs
doc.tables (recurse into cell.paragraphs and cell.tables)
doc.sections (iterate section.header.paragraphs and section.footer.paragraphs)