Repository Intelligence Tools - Analyze codebases with LLM assistance

Repo Intel is a comprehensive toolkit for analyzing repositories, generating documentation, and performing AI-assisted code reviews. It provides three main tools:

• Git Diff Analyzer - Compare branches with LLM-powered code reviews
• Markdown Bundler - Combine code files into LLM-friendly markdown bundles
• Glue Documenter - Generate documentation for AWS Glue databases

pip install repo-intel

# Analyze differences between branches
repo-intel diff-analyze main staging

# Create a markdown bundle of your codebase
repo-intel markdown-bundle src/ -o codebase.md

# Document AWS Glue databases
repo-intel glue-document -d my_database

# List available LLM providers
repo-intel list-providers

• File-by-file analysis that breaks down large diffs into manageable chunks
• LLM integration with OpenAI, Anthropic, and local models
• Risk assessment with automatic priority ranking
• Smart filtering to skip oversized or binary files
• Comprehensive reporting with both summary and detailed analysis

• Flexible file inclusion with configurable extensions
• Smart exclusion patterns for common directories (node_modules, .git, etc.)
• Organized output with table of contents and proper formatting
• Markdown-only mode for documentation bundling

• Complete AWS Glue documentation for databases and tables
• Schema documentation with column details and types
• Metadata extraction including creation dates and parameters
• Flexible filtering to exclude specific databases or tables

Repo Intel uses environment variables for configuration. Create a .env file or set these in your environment:

# Choose your LLM provider
LLM_PROVIDER=openai  # 'openai', 'anthropic', 'local', or leave empty for auto-select

# OpenAI settings
OPENAI_API_KEY=your_openai_api_key
OPENAI_MODEL=gpt-4
OPENAI_TEMPERATURE=0.3

# Anthropic settings  
ANTHROPIC_API_KEY=your_anthropic_api_key
ANTHROPIC_MODEL=claude-3-sonnet-20240229

# Local LLM settings (Ollama, etc.)
LOCAL_LLM_BASE_URL=http://localhost:11434
LOCAL_LLM_MODEL=codellama

AWS_REGION=us-west-2
AWS_PROFILE=default

# Default output directory
OUTPUT_DEFAULT_DIR=repo_intel_output

# Maximum file size for LLM analysis (bytes)
LLM_DEFAULT_MAX_FILE_SIZE=250000

# Enable verbose logging
OUTPUT_VERBOSE=true

# Basic branch comparison
repo-intel diff-analyze main feature/new-api

# Custom output directory and file size limit
repo-intel diff-analyze main staging \
    --output-dir my_review \
    --max-file-size 500000

# Skip LLM analysis for faster processing
repo-intel diff-analyze main staging --no-llm

# Force specific LLM provider
repo-intel diff-analyze main staging --llm-provider anthropic

# Bundle all code files
repo-intel markdown-bundle src/

# Bundle only markdown files
repo-intel markdown-bundle docs/ --markdown-only

# Exclude additional patterns
repo-intel markdown-bundle . --exclude __pycache__ .pytest_cache

# Custom output file
repo-intel markdown-bundle src/ -o my_codebase.md

# Document all databases
repo-intel glue-document

# Document specific database
repo-intel glue-document -d my_database

# Use specific AWS profile and region
repo-intel glue-document --profile prod --region us-east-1

# Exclude specific databases or tables
repo-intel glue-document \
    --exclude-databases temp_db test_db \
    --exclude-tables temp_table

repo-intel/
├── src/repo_intel/
│   ├── __init__.py          # Package version and metadata
│   ├── cli.py               # Main CLI interface
│   ├── diff.py              # Git diff analyzer
│   ├── llm.py               # LLM provider integrations
│   ├── markdown_bundle.py   # Markdown bundler
│   ├── glue_bundle.py       # AWS Glue documenter
│   └── settings.py          # Configuration management
├── tests/                   # Test suite
├── docs/                    # Documentation
├── setup.py                 # Package setup
├── pyproject.toml          # Modern Python packaging
├── requirements.txt        # Dependencies
└── README.md               # This file

1. Get an API key from OpenAI
2. Set OPENAI_API_KEY environment variable
3. Choose your model (gpt-4, gpt-3.5-turbo, etc.)

1. Get an API key from Anthropic
2. Set ANTHROPIC_API_KEY environment variable
3. Use Claude models for analysis

1. Install Ollama
2. Pull a code model: ollama pull codellama
3. Start service: ollama serve
4. Configure LOCAL_LLM_BASE_URL and LOCAL_LLM_MODEL

repo_intel_output/
├── README.md                # Summary with risk-ranked files
├── summary.json            # Machine-readable summary
├── detailed_analysis.json  # Complete analysis data
└── files/                  # Individual file reports
    ├── src_main.py.md
    ├── api_routes.py.md
    └── config_settings.py.md

Files are automatically prioritized:

• Critical (8-10): Core changes requiring immediate attention
• High (6-7): Important changes needing careful review
• Medium (4-5): Standard changes requiring normal review
• Low (1-3): Minor changes needing quick review

Contributions are welcome! Please feel free to submit a Pull Request.

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

• Issues: GitHub Issues
• Documentation: GitHub README
• Discussions: GitHub Discussions

• Support for more LLM providers (HuggingFace, local models)
• Integration with popular code review tools
• Advanced filtering and configuration options
• Web UI for report viewing
• CI/CD integration examples
• Plugin system for custom analyzers

Made with ❤️ for developers who love clean, well-analyzed code