CodeLoom

CodeLoom is an AI-powered tool for analyzing and optimizing your codebase structure. Leveraging Google Gemini models, known for their long context capabilities, it provides deep insights into your code organization and offers targeted suggestions for improving critical files.

Installation and Usage

You can use CodeLoom either by installing it globally or by running it directly with npx.

Global Installation

1. Install CodeLoom globally using npm:

   npm install -g @bilalpm/codeloom

2. Use CodeLoom in any project:

   cd path/to/your/project
   codeloom <directories...>

Using npx (No Installation Required)

If you prefer not to install CodeLoom globally, you can use npx to run it directly:

npx @bilalpm/codeloom <directories...>

Setup

Before using CodeLoom, create a .env.local file in your project root with your Gemini API key:

GEMINI_API_KEY=your_api_key_here

Replace your_api_key_here with your actual Gemini API key.

Usage Examples

Analyze specific directories:

codeloom app lib  # If installed globally
npx @bilalpm/codeloom app lib  # Using npx

Use command-line options:

codeloom app -m 5 -l 1000 -o ./codeloom-analysis -n gemini-1.5-flash-latest  # If installed globally
npx @bilalpm/codeloom app components -m 5 -l 1000 -o ./codeloom-analysis -n gemini-1.5-flash-latest  # Using npx

This command will:

• Analyze the app and components directories
• Consider up to 5 critical files for optimization
• Include files up to 1000 lines long
• Output results to ./codeloom-analysis directory
• Use the gemini-1.5-flash-latest model for analysis

Command-line Options

• -m, --max-critical-files <number>: Maximum number of critical files to analyze (default: 3)
• -o, --output-dir <path>: Custom output directory for CodeLoom files (default: "./codeloom_out")
• -l, --max-lines <number>: Maximum number of lines per file to include in analysis (default: 500)
• -n, --model-name <name>: Model name to use for analysis (default: "gemini-1.5-flash-latest")

Output

After running CodeLoom, you'll find a new output directory (default: codeloom_out) in your project root. It will contain:

• codeloom-map.json: A JSON representation of your codebase structure
• codeloom-output.txt: Concatenated content of all analyzed files
• organization-suggestions.json: AI-generated suggestions for codebase organization
• critical-files-suggestions.json: List of critical files identified by AI
• optimized_files/: Directory containing optimized versions of critical files

CodeLoom will provide optimization suggestions for the identified critical files, saving the optimized versions in the optimized_files/ directory while preserving the original file structure.

Note

CodeLoom is an AI-powered tool and its suggestions should be reviewed by a developer before implementation. Always back up your code before applying any automated changes.

License

MIT

Author

Bilal Tahir