drommage 1.0.4

AI-powered git commit analysis tool for developers and teams

🚀 DRommage - Intelligent Documentation Version Control

Where git meets AI to understand not just what changed, but why it matters

🎯 What is DRommage?

DRommage is a revolutionary documentation versioning system that combines git-style region tracking with LLM-powered semantic analysis. It doesn't just show you diffs - it explains them, analyzes their impact, and provides actionable insights.

✨ Key Features

🧠 Intelligent Diff Analysis

• Semantic Understanding: Goes beyond line-by-line diffs to understand meaning
• Multi-Level Analysis: Brief summaries to deep technical breakdowns
• Risk Assessment: Identifies potential issues and breaking changes
• Smart Recommendations: Suggests next steps and improvements

🎨 Beautiful Terminal UI

• Golden Ratio Layout: Aesthetically pleasing proportions
• Unicode Box Drawing: Modern, clean interface design
• Real-time Status Updates: Never wonder what's happening
• Color-Coded Regions: Instantly see volatile vs stable areas

⚡ Performance & Efficiency

• SQLite Caching: Lightning-fast repeated analyses
• Local LLM Support: Privacy-first with Ollama integration
• Region Tracking: Efficiently monitors document evolution
• Incremental Analysis: Only analyzes what changed

🛠️ Installation

Prerequisites

# Python 3.8+
python --version

# Ollama for local LLM (recommended)
# macOS:
brew install ollama

# Linux:
curl -fsSL https://ollama.ai/install.sh | sh

# Pull the Mistral model
ollama pull mistral

Quick Start

# Clone the repository
git clone [e-mail]:yoursteacup/Drommage.git
cd Drommage

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Run DRommage
python drommage.py

🎮 Usage

Keyboard Controls

Key	Action
↑↓	Navigate commits
SPACE	Toggle analysis mode (PAT → BRIEF → DEEP)
D	Run analysis for current mode
Q/E	Flip through pages
R/F	Scroll analysis results
←→	Scroll diff view
H	Help
Q	Quit

Analysis Modes

🔍 PAT Analysis (Pattern)

Quick pattern-based analysis without LLM - works offline, shows file types and basic metrics.

📝 BRIEF Analysis (AI-powered)

Quick AI summary of changes - perfect for scanning through commits.

🔬 DEEP Analysis (AI-powered)

Comprehensive analysis including:

• Change type classification (feature/fix/docs/etc.)
• Impact assessment
• Detailed explanation
• Potential risks
• Actionable recommendations

🏗️ Architecture

DRommage/
├── README.md           # Comprehensive documentation
├── drommage.py        # Main entry point
├── requirements.txt   # Dependencies (none!)
├── drommage/          # Core package
│   ├── __init__.py
│   └── core/
│       ├── diff_tracker.py    # Git-style region tracking
│       ├── llm_analyzer.py    # LLM integration & analysis
│       ├── region_analyzer.py # Region volatility tracking
│       └── interface.py       # Terminal UI implementation
├── test_docs/         # Sample documents
├── test_scripts/      # Helper scripts
└── old_refs/          # Version history

Core Components

1. DiffTracker: Tracks document regions through versions using semantic hashing
2. LLMAnalyzer: Interfaces with Ollama for intelligent analysis
3. RegionAnalyzer: Identifies volatile vs stable document sections
4. Interface: Beautiful terminal interface with real-time updates

🔬 How It Works

1. Region Detection: Documents are parsed into semantic regions
2. Change Tracking: Each region gets a unique hash and is tracked across versions
3. Smart Analysis: When you request analysis, the LLM examines:
   • What changed (text diff)
   • Why it changed (semantic analysis)
   • What it means (impact assessment)
4. Caching: Results are stored for instant retrieval

💡 Use Cases

• API Documentation: Track breaking changes and compatibility
• Technical Specs: Monitor requirement evolution
• Knowledge Bases: Understand content drift over time
• Compliance Docs: Ensure critical sections remain stable
• Team Documentation: See who changed what and why

🚦 Status Indicators

During analysis, you'll see:

• 📊 Analyzing: X lines, Y chars diff - Diff metrics
• 📝 Prompt size: X chars, Level: brief/detailed - LLM prompt info
• 🤖 Starting LLM inference... - Processing
• ✅ Analysis complete (X.Xs) - Done with timing
• 📦 Using cached analysis - Retrieved from cache

🎨 Visual Design

The interface uses carefully chosen visual elements:

• Colors: 11-color palette for different change types
• Icons: Semantic icons (🚀 feature, 🐛 bugfix, 📝 docs, etc.)
• Layout: Golden ratio (0.382) for panel sizing
• Borders: Unicode box drawing for clean lines

🔧 Configuration

Using Different LLM Models

# In main_v8.py, modify:
llm = LLMAnalyzer(model="mistral:latest")  # Change model here

Adjusting Cache Location

# Default: .llm_cache/
llm = LLMAnalyzer(cache_dir="custom_cache_path")

🐛 Troubleshooting

Ollama Not Found

# Check installation
ollama list

# If not installed, see Prerequisites section

Analysis Seems Stuck

• Check status bar for progress updates
• Default timeout is 30 seconds
• Press Q to quit if needed

Cache Issues

# Clear cache if corrupted
rm -rf .llm_cache/

🚀 Future Roadmap

• Cloud LLM fallback (OpenAI/Claude API)
• Web UI version
• Collaborative features
• Export to CHANGELOG.md
• Git hooks integration
• VS Code extension

🤝 Contributing

We welcome contributions! Whether it's:

• Bug reports
• Feature requests
• Code improvements
• Documentation updates

Feel free to open an issue or submit a PR.

📜 License

MIT License - See LICENSE file for details

🙏 Acknowledgments

• Built with Anthropic's Claude
• Powered by Ollama and Mistral
• Inspired by the need for smarter documentation tools

---

Remember: DRommage doesn't just track changes - it understands them. 🧠✨

Happy documenting!