StudyGenerator Agent

Accessible study material generator for students with learning differences. Uses Claude vision to read scanned textbooks and generates optimized PDFs.

What This Agent Does

Reads scanned textbooks using Claude vision built-in
Extracts and simplifies content for students with learning differences
Generates accessible PDFs with:
- OpenDyslexic font
- All uppercase text
- Simplified concepts
- Mind maps
- Step-by-step exercise guides

Supported Profiles

This agent generates material for dyslexia profile:

Large font (14pt+)
High contrast (black on white/light gray)
Uppercase text
Short sentences
Visual concepts (mind maps)
Step-by-step guided exercises

Workflow

1. Receive Input

Required input:
- Scanned textbook PDF (local path)
- Topic to cover
- Specific pages (optional)
- Exercises to include (optional)

2. Analyze with Vision

# Claude reads PDF page by page using vision
# Extracts:
# - Titles and subtitles
# - Key concepts
# - Formulas
# - Definitions
# - Exercises

3. Structure Content

For each topic, create:

Cover page - Title, student name, index
Overview map - All topics connected
For each section:
- Simplified summary in concept boxes
- Practical examples
- Mind map
Exercises:
- Formula reference sheet
- Step-by-step guides

4. Generate Python Script

Create/modify script in testingcase/ based on existing templates:

# Base template: genera_pdf_mario_light.py
# Uses ReportLab with:
# - OpenDyslexic font (if available)
# - Concept boxes
# - Mind maps
# - Exercise guides

5. Execute and Verify

# Run script
python testingcase/genera_pdf_{topic}.py

# Verify output
open ~/Downloads/{OUTPUT_FILE}.pdf

PDF Output Structure

1. COVER PAGE
   - Subject title
   - Topic
   - "Study material for [Student Name]"
   - Topic index

2. OVERVIEW MAP
   - Central node = main theme
   - Branches = sub-topics
   - Key formulas visible

3. FOR EACH TOPIC:
   a. Section title (large font, underlined)
   b. Concept boxes:
      - Box title
      - Simple bullet points (max 4 per box)
   c. Practical example (dashed box)
   d. Topic mind map

4. FORMULA REFERENCE
   - All formulas large
   - Variants (how to find X, how to find K)

5. EXERCISE GUIDES
   - Problem text
   - "USE THE FORMULA: ..."
   - STEP 1, 2, 3, 4
   - Final answer box

Python Script Template

#!/usr/bin/env python3
"""
Generate accessible PDF for [STUDENT] - [TOPIC]
"""

from reportlab.lib.pagesizes import A4
from reportlab.lib.units import cm
from reportlab.pdfgen import canvas
from reportlab.lib.colors import black, white, HexColor
import os

# Configuration
OUTPUT_PDF = "~/Downloads/[NAME]_[TOPIC].pdf"
STUDENT = "[NAME]"
TOPIC = "[TOPIC]"

# DSA-friendly colors
LIGHT_GRAY = HexColor("#F5F5F5")
BLACK = black
WHITE = white

# Font (fallback to Helvetica if OpenDyslexic unavailable)
FONT_DIR = os.path.join(os.path.dirname(__file__), "fonts")
try:
    from reportlab.pdfbase import pdfmetrics
    from reportlab.pdfbase.ttfonts import TTFont
    pdfmetrics.registerFont(TTFont('OpenDyslexic',
        os.path.join(FONT_DIR, 'OpenDyslexic-Regular.ttf')))
    FONT_NORMAL = 'OpenDyslexic'
    FONT_BOLD = 'OpenDyslexic-Bold'
except:
    FONT_NORMAL = 'Helvetica'
    FONT_BOLD = 'Helvetica-Bold'

# ... helper functions (see genera_pdf_mario_light.py) ...

def main():
    c = canvas.Canvas(OUTPUT_PDF, pagesize=A4)
    # Generate content
    c.save()
    print(f"PDF created: {OUTPUT_PDF}")

if __name__ == "__main__":
    main()

Available Helper Functions

Scripts in testingcase/ provide these reusable functions:

| Function | Description | |----------|-------------| | disegna_titolo_sezione(c, title, y) | Large title with underline | | disegna_box_concetto(c, title, contents, y) | Gray box with bullet points | | disegna_esempio(c, text, y) | Dashed box for examples | | calcola_punto_bordo_ellisse(...) | For mind map connections | | calcola_punto_bordo_box(...) | For mind map connections |

Invocation

From CLI

# Generate PDF for new topic
claude "Generate accessible PDF for Mario on [TOPIC]
        using pages [N-M] from textbook [PATH]"

Full Example

claude "Generate accessible PDF for Mario on KINEMATICS.
        Textbook: ~/Downloads/physics_ch5.pdf
        Pages: 1-4
        Exercises: 81, 83, 85

        Required structure:
        1. Average velocity
        2. Uniform rectilinear motion
        3. Acceleration"

Generation Checklist

Before delivering the PDF, verify:

[ ] Readable font (OpenDyslexic or large Helvetica)
[ ] All text UPPERCASE
[ ] Simplified concepts (max 4 bullets per box)
[ ] Overview map present
[ ] Mind map for each topic
[ ] Large, clear formulas
[ ] Step-by-step exercise guides
[ ] No walls of text
[ ] High contrast (black on white/gray)

Related Files

testingcase/genera_pdf_mario.py - Full template with book pages
testingcase/genera_pdf_mario_light.py - Template without book pages
testingcase/fonts/OpenDyslexic-Regular.ttf - DSA font regular
testingcase/fonts/OpenDyslexic-Bold.ttf - DSA font bold
testingcase/fonts/OpenDyslexic-Italic.ttf - DSA font italic
testingcase/fonts/AtkinsonHyperlegible-*.ttf - Alternative font (visual impairment)
testingcase/genera_pdf_atmosfera.py - Example for another topic

Limitations

Dyslexia profile only - Other profiles (ADHD, visual, etc.) require MirrorBuddy integration
Static PDF - Not interactive (no audio, no embedded quizzes)
Requires Python - To run generated scripts
Manual vision - Claude reads pages, not automatic OCR

Future Development

Component 2 (MirrorBuddy Integration) will add:

7 complete accessibility profiles
Azure OpenAI GPT-4o for automatic vision
@react-pdf/renderer (solves OpenDyslexic font in browser)
Study Kit integration
Full learning path export

Studygenerator Agent