Scriptoria Project V2

A modular, AI-driven pipeline for intelligent document processing—from raw URLs to polished content.

---

🎯 Overview

Scriptoria Project is an advanced AI-powered content processing system that transforms web content through intelligent scraping, reinforcement learning, and multi-agent AI collaboration. The system processes URLs through a sophisticated pipeline that includes content extraction, AI analysis, rewriting, and human-in-the-loop feedback.

🎥 Live Demo & UI Walkthrough

Main Interface	Content Processing
Enhanced Output	Multi-modal Input

---

✨ Key Features

🤖 AI-Powered Processing

• Intelligent Content Analysis: AI-driven review and scoring of web content quality
• Multi-Agent Collaboration: Writer and reviewer agents powered by Google Gemini
• Reinforcement Learning: Adaptive search and reward scoring for optimal content discovery

🎯 Multi-Modal Interface

• Streamlit Web UI: Interactive interface with real-time processing visualization
• FastAPI Backend: RESTful API for integration with other applications
• CLI Orchestrator: Command-line interface for automated workflows
• Voice Integration: speech-to-text via Vosk with audio processing
• Speaking Agent: text-to-speech on AI Written summary page

🔧 Advanced Architecture

• Modular Pipeline: Extensible components for web scraping, AI processing, and content enhancement
• Vector Storage: Semantic search and retrieval using ChromaDB
• Cross-Platform: Native support for Windows, Linux, and macOS
• GPU Optimization: Optional GPU acceleration with CPU fallback

---

🏗️ System Architecture

graph TD
    A[URL, Name, RLQuery Input] --> B[Web Scraping & Screenshot]
    B --> C[RL Search & Scoring]
    C --> D[AI Content Analysis/Review]
    D --> E[Content Quality Scoring]
    E --> F[AI Rewriting]
    F --> G[Human Feedback]
    G --> H{Feedback Type}
    H --> I[Text Input]
    H --> J[Audio Input]
    H --> K[No Input]
    I --> D
    J --> D
    K --> L[Final Output]
    L --> M[Restart Workflow]
    L --> N[Vector Storage Deletion]
    M --> A

Loading

---

🚀 Quick Start

Prerequisites

• Python 3.11 or higher
• Google Gemini API Key (Get free API key)

Installation Options

🐍 Option 1: Conda Environment (Recommended)

# Create environment from YAML
conda env create -f environment.yml
conda activate scriptenv

# Install Playwright browsers
playwright install

🛠️ Option 2: Virtual Environment and Install Dependencies

# Create virtual environment
python -m venv scriptenv

# Activate (Windows PowerShell)
.\scriptenv\Scripts\Activate.ps1

# If permission Error:
Set-ExecutionPolicy -ExecutionPolicy Bypass -Scope Process -Force

# Install Python packages
pip install -r requirements.txt

# Install Playwright browsers
playwright install

🔑 Set API Key

# PowerShell
$env:GEMINI_API_KEY="your-api-key-here"

# bash/zsh
export GEMINI_API_KEY="your-api-key-here"

---

🎮 Usage Modes

🌐 Streamlit UI (Recommended for Beginners)

streamlit run src/streamlit_app.py

Features:

• Interactive web interface
• Real-time processing visualization
• Audio input and output support
• Step-by-step workflow guidance

🔌 FastAPI Backend

uvicorn src.api_server:app --reload

API Endpoints:

• GET / - API documentation
• GET /write?url=link - Content writing endpoint
• GET /review?url=link - Content review endpoint

💻 CLI Orchestrator

python src/main.py

Features:

• Lightweight command-line interface
• Automated batch processing
• Integration with existing workflows

---

📁 Project Structure

Scriptoria-Project/
├── agents/                   # AI Agent Modules
│   ├── ai_writer.py          # Content generation agent
│   ├── ai_reviewer.py        # Quality assessment agent
│   └── voice_api.py          # Speech processing
├── chromadb/                 # Vector Database Storage
├── data/
│   ├── demo/                 # Demonstration screenshots
│   ├── logo/                 # Brand assets
│   ├── model/                # Vosk speech model
│   ├── raw_content/          # Original content storage
│   ├── processed_content/    # Enhanced content storage
│   └── screenshots/          # UI documentation
├── src/                      # Core Application
│   ├── api_server.py         # FastAPI backend
│   ├── main.py               # CLI entry point
│   ├── rl_reward.py          # RL search result reward function
│   ├── rl_search.py          # Intelligent search
│   ├── scraper.py            # Web content extraction
│   ├── streamlit_app.py      # Primary GUI Application
│   └── versioning.py         # Content version management
├── .vscode/                  # Editor settings
├── example-urls.txt          # Sample URLs & friendly names
├── requirements.txt          # Python dependencies
├── terminal-commands.txt     # Step-by-step guide to run it in CLI, Streamlit and API mode
├── enviroment.yml            # Enviroment Config Setup by "conda".
├── output.mp3                # Output MP3 generated from streamlit_app.py and acts as a record of last webpage audio
└── README.md                 # Instructions file

---

🎯 Workflow Demonstration

Step 1: Input & Configuration

Features:

• URL input with validation
• Content naming and categorization
• Reinforcement learning query configuration
• Example URL integration

Step 2: Content Acquisition

Process:

• Intelligent web scraping with HTML cleaning
• Screenshot capture for visual reference
• Content structure analysis

Step 3: AI Analysis & Scoring

Analysis Includes:

• Content quality assessment
• Readability scoring
• Improvement recommendations
• Text-to-speech audio generation

Step 4: Content Enhancement

Enhancement Features:

• AI-powered content rewriting
• Quality improvement tracking
• Style and tone optimization
• Multi-format output support

Step 5: Human-in-the-Loop

Text Feedback

Audio Feedback

Feedback Options:

• Text-based suggestions and edits
• Voice feedback with speech-to-text
• Iterative improvement cycles
• Quality validation

Step 6: Final Output & Options

Output Features:

• Version management
• Process restart capability
• Export functionality

---

🌐 Example Content Sources

The system works with various web content types. Example URLs include:

Content Title	Example URL	Content Type	Use Case
Joy of Discipline	library.acropolis.org	Philosophical Essay	Personal development insights
Gates of Morning	Wikisource	Literature	Classic text modernization
Born or Built Smart	Psychology Today	Article	Intelligence and effort analysis
Sufficient Reason	Stanford Encyclopedia	Academic	Conceptual deep dive
Infinity's Existence	Scientific American	Research	Abstract theory accessibility

Sample URLs File: example-urls.txt contains curated starting points.

---

🖥️ CLI Mode Examples

CLI Interface Top Output

CLI Interface Bottom Output

CLI Advantages:

• Scriptable and automatable
• Resource-efficient operation
• Batch processing capabilities
• Integration with CI/CD pipelines

---

🔧 Advanced Configuration

Environment Variables

# Required
GEMINI_API_KEY="your-gemini-api-key"

Custom Content Processing

1. src/rl_reward.py for custom scoring algorithms
2. agents/ai_writer.py for writing style customization
3. agents/ai_reviewer.py for reviewing style customization
4. src/versioning.py for different aspects of versioning.

---

🚀 Performance Optimization

GPU Acceleration

The system automatically detects and utilizes GPU resources when available. Key optimizations include:

• TensorFlow GPU support for ML operations
• ONNX Runtime for model inference acceleration
• Parallel processing for multi-document handling

Memory Management

• Chunked processing for large documents
• Efficient vector storage with ChromaDB
• Automatic cache management

---

🤝 Contributing

We welcome contributions! Please see our development guidelines:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

---

🆘 Support & Troubleshooting

Common Issues

API Key Problems:

# Verify API key is set
echo $GEMINI_API_KEY  # Linux/Mac
echo $env:GEMINI_API_KEY  # Windows PowerShell

Playwright Installation:

# Reinstall browsers if needed
playwright install

Dependency Conflicts:

# Fresh environment setup
conda env remove -n scriptenv
conda env create -f environment.yml

---

🎊 Acknowledgments

• Google Gemini for AI capabilities
• Vosk for speech-to-text functionality
• Chromadb for vector storage solutions
• Streamlit for interactive UI components
• Playwright for robust web scraping

---

🛠️ Made with Precision & 🧠

✨ Kratu Gautam ✨

---

🎯 Architect of agentic RL workflows, reproducible environments, and browser-audible AI interfaces.
💡 Driven by clarity, modularity, and a passion for empowering teams through automation and documentation.

🔗 GitHub: Kratugautam99
📘 Project: Scriptoria – AI-Driven Content Processing

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---