academic-asset-generator

name

academic-asset-generator

description

Transform raw academic ideas into professional four-asset packages (DOCX, PDF, PPTX, HTML poster) within a single session. Use when students or researchers need to create comprehensive academic deliverables from initial concepts, including research papers, presentations, conference posters, or thesis documents. Automates format layout and ensures consistency across all outputs through structured metadata as single source of truth.

Academic Asset Generator

Overview

This skill transforms raw academic ideas into production-ready asset packages that meet industrial and academic standards. By automating the "format layout" work, it eliminates low-value repetitive labor and delivers professional outputs.

Core Concept: Single Source of Truth

All assets are generated from Structured Metadata - a JSON document that serves as the canonical source for all content. This ensures perfect consistency: the first chapter in the DOCX matches the first slide in the PPTX, which matches the first section in the HTML poster.

Raw Idea → Structured Metadata → [DOCX + PDF + PPTX + HTML Poster]

Complete Workflow

Step 1: Understand the Requirement

When a user presents a raw idea, ask clarifying questions to understand:

1. Type of deliverable: Research paper, presentation, poster, thesis, report?
2. Target audience: Conference, journal, class assignment, general audience?
3. Content scope: Main topics, sections, key messages
4. Visual elements: Figures, tables, diagrams needed?
5. Style preferences: Formal/informal, colorful/minimal, technical level

Example questions:

• "What's the main message you want to convey?"
• "Who is your target audience?"
• "Do you have any figures, data, or images to include?"
• "What format is this for - a conference, journal, or class?"
• "Are there specific style requirements or templates to follow?"

Step 2: Generate Structured Metadata

Create a JSON file containing all content in structured form. This becomes the Single Source of Truth for all subsequent assets.

Read the schema: See metadata-schema.md for complete JSON structure and examples.

Key principles:

• All content goes into the metadata first
• Use semantic structure (sections, subsections)
• Include all metadata (authors, dates, affiliations)
• Reference external files (images, data) by path
• Validate IDs are unique and hierarchy is logical

Save as: metadata.json

Example minimal metadata:

{
  "project": {
    "title": "Machine Learning for Climate Prediction",
    "subtitle": "A Novel Deep Learning Approach",
    "authors": [
      {
        "name": "Jane Smith",
        "affiliation": "University of Example",
        "email": "[e-mail]"
      }
    ],
    "date": "2026-02-03",
    "keywords": ["machine learning", "climate", "prediction"],
    "abstract": "This research presents a novel deep learning approach for improving climate prediction accuracy by 15%."
  },
  "sections": [
    {
      "id": "intro",
      "title": "Introduction",
      "level": 1,
      "content": [
        {
          "type": "paragraph",
          "text": "Climate prediction remains one of the most challenging problems in environmental science."
        },
        {
          "type": "list",
          "items": [
            "Current models have 70% accuracy",
            "Deep learning shows promise",
            "Our approach achieves 85% accuracy"
          ]
        }
      ]
    },
    {
      "id": "methods",
      "title": "Methodology",
      "level": 1,
      "content": [
        {
          "type": "paragraph",
          "text": "We developed a transformer-based architecture trained on 20 years of climate data."
        }
      ]
    },
    {
      "id": "results",
      "title": "Results",
      "level": 1,
      "content": [
        {
          "type": "paragraph",
          "text": "Our model demonstrates a 15% improvement over baseline methods."
        }
      ]
    },
    {
      "id": "conclusion",
      "title": "Conclusions",
      "level": 1,
      "content": [
        {
          "type": "paragraph",
          "text": "This work demonstrates the potential of deep learning for climate science."
        }
      ]
    }
  ]
}

Step 3: Generate DOCX Document

Create a professional Word document from the metadata.

Read the guide: See docx-generation.md for complete implementation details.

Requirements:

• Standard academic formatting (margins, fonts, spacing)
• Hierarchical section structure
• Professional typography
• Proper heading levels

Process:

1. Read metadata.json
2. Create Document using docx library
3. Apply standard academic styling
4. Export to outputs/document.docx

Standard format:

• Font: Times New Roman, 12pt
• Line spacing: 1.5 or double
• Margins: 1 inch all sides
• Headings: Bold, hierarchical sizing

Step 4: Generate PDF from DOCX

Convert the Word document to PDF for final distribution.

soffice --headless --convert-to pdf outputs/document.docx --outdir outputs

Result: outputs/document.pdf

Step 5: Generate PPTX Presentation

Create a professional PowerPoint presentation from the metadata.

Read the guide: See pptx-generation.md for complete implementation details.

Requirements:

• Title slide with author information
• Section divider slides for major sections
• Content slides with clear hierarchy
• Consistent color scheme and typography
• Readable from distance (18pt+ body, 28pt+ titles)

Process:

1. Read metadata.json
2. Select appropriate color palette based on subject matter
3. Create slides using PptxGenJS
4. Apply consistent layout patterns
5. Export to outputs/presentation.pptx

Layout patterns:

• Title slide: Centered, bold, large text
• Section dividers: Full-screen section titles
• Content slides: Title + bullet points or title + image
• Two-column layout: Text left, visual right

Step 6: Generate HTML Poster

Create an interactive, print-ready HTML research poster.

Read the guide: See html-poster-generation.md for complete implementation details.

Requirements:

• Standard poster size (A0: 841mm × 1189mm or custom)
• Multi-column grid layout (typically 3 columns)
• High contrast and readability
• Print-ready with @page CSS
• Optional: SVG graphics, QR codes, interactive elements

Process:

1. Read metadata.json
2. Create HTML with embedded CSS
3. Structure content in column grid
4. Apply poster-specific styling (large fonts, high contrast)
5. Save to outputs/poster.html

Convert to PDF for printing:

npx playwright screenshot outputs/poster.html outputs/poster.pdf --full-page

Step 7: Package and Deliver

Organize all outputs in the outputs/ directory:

outputs/
├── document.docx
├── document.pdf
├── presentation.pptx
└── poster.html
└── poster.pdf (if printed)

Final checklist:

• All files generated successfully
• Content is consistent across all formats
• Formatting meets academic standards
• Files are in outputs/ directory
• No errors or missing content

Leveraging Existing Skills

This skill builds upon and integrates with existing skills in the sandbox:

DOCX Skill

Use the docx skill for advanced document operations:

• Tracked changes for collaborative editing
• Complex table formatting
• Advanced styling and templates

PPTX Skill

Use the pptx skill for advanced presentation features:

• Using existing templates
• Advanced chart generation
• Thumbnail validation

PDF Skill

Use the pdf skill for PDF manipulation:

• Merging multiple PDFs
• Adding watermarks
• Form filling

Generate-Image Skill

Use the generate-image skill for creating custom graphics:

• Diagrams and illustrations
• Data visualizations
• Icons and graphics

Integration approach:

1. Generate core assets with this skill
2. Use specialized skills for advanced features
3. Maintain metadata.json as single source of truth

Asset Type Decision Guide

Choose the right combination based on use case:

Conference Paper Submission:

• ✓ DOCX (for review/editing)
• ✓ PDF (for submission)
• Optional: PPTX (for presentation)

Conference Presentation:

• ✓ PPTX (primary deliverable)
• ✓ PDF (handout version)
• Optional: HTML Poster (conference poster session)

Research Poster Session:

• ✓ HTML Poster → PDF (for printing)
• Optional: DOCX (extended abstract)

Class Assignment:

• ✓ DOCX (for review/comments)
• ✓ PDF (for final submission)
• Optional: PPTX (for presentation)

Thesis/Dissertation:

• ✓ DOCX (for advisor review)
• ✓ PDF (for final submission)
• ✓ PPTX (for defense)

Best Practices

Content Organization

1. Logical flow: Introduction → Methods → Results → Conclusions
2. Section hierarchy: Use consistent heading levels
3. One idea per section: Keep sections focused
4. Visual balance: Distribute content evenly

Consistency

1. Single source of truth: Always update metadata.json first
2. Regenerate all assets: After metadata changes, regenerate everything
3. Version control: Track metadata.json changes
4. Naming conventions: Use consistent file naming

Quality Control

1. Validate metadata: Check for missing fields, broken references
2. Test generation: Generate all assets before finalizing
3. Cross-check content: Verify consistency across formats
4. Review outputs: Check formatting, typography, layout

Performance

1. Reuse components: Create helper functions for common patterns
2. Batch processing: Generate all assets in one script
3. Cache images: Don't regenerate unchanged images
4. Minimize dependencies: Use standard libraries when possible

Example Use Cases

Use Case 1: Conference Paper

User: "I need to write a conference paper about my research on neural networks for image recognition."

Process:
1. Ask about: target conference, paper length, key findings, figures available
2. Generate metadata.json with abstract, introduction, methods, results, conclusion
3. Generate DOCX with standard conference format
4. Generate PDF for submission
5. Generate PPTX for presentation (if accepted)

Use Case 2: Research Poster

User: "I need a poster for the science fair about renewable energy."

Process:
1. Ask about: key message, target audience, data/images available
2. Generate metadata.json with sections: problem, solution, results, impact
3. Generate HTML poster with 3-column layout
4. Convert to PDF for printing

Use Case 3: Thesis Chapter

User: "Help me format Chapter 3 of my thesis on quantum computing."

Process:
1. Ask about: thesis style guidelines, content structure, figures needed
2. Generate metadata.json with hierarchical sections
3. Generate DOCX with thesis formatting
4. Generate PDF for committee review

Troubleshooting

Issue: Content doesn't fit in layout

Solution: Reduce text length, split into multiple slides/sections, or adjust font sizes

Issue: Inconsistent styling across assets

Solution: Verify all assets are generated from same metadata.json, regenerate all

Issue: Images not displaying

Solution: Check image paths in metadata.json, ensure files exist, use absolute or relative paths consistently

Issue: PDF conversion fails

Solution: Ensure LibreOffice is installed, check file permissions, verify DOCX is valid

Dependencies

Required tools (should be available in sandbox):

• Node.js libraries:

  • docx - Word document generation
  • pptxgenjs - PowerPoint generation
  • playwright - HTML to PDF conversion

• System tools:

  • libreoffice - DOCX to PDF conversion
  • wkhtmltopdf - Alternative HTML to PDF

• Optional:

  • sharp - Image processing
  • chartjs - Chart generation

Skill Integration

This skill is designed to work seamlessly with other skills:

• docx: Advanced document manipulation
• pptx: Template-based presentations
• pdf: PDF manipulation and forms
• generate-image: Custom graphics generation
• frontend-design: Custom HTML/CSS designs

When specialized features are needed, delegate to the appropriate skill while maintaining the metadata.json as the single source of truth.