llm-codebase-context

LLM Codebase Context

A command-line tool to prepare your codebase for submission to Large Language Models (LLMs) like ChatGPT or Claude. It creates a formatted context file that includes your source code while intelligently filtering out binary files, large files, and commonly ignored directories.

Why Use This?

• Quickly create LLM-ready snapshots of your codebase
• Smart filtering removes irrelevant files (node_modules, binaries, build artifacts)
• Avoids common issues like token limits by excluding overly large files
• Structured XML output format is ideal for LLM context

Installation

# Install globally
npm install -g llm-codebase-context

# Or run with npx without installing
npx llm-codebase-context <project_root>

Usage

Basic Usage

# Process the current directory and output to structure.txt
codebase-context .

# Process a specific project and specify output file
codebase-context /path/to/project output.txt

# Process only a subdirectory of your project
codebase-context /path/to/project output.txt src

Advanced Usage

# Add custom ignore patterns
codebase-context . output.txt . "tests,fixtures,*.json"

# Process a specific subdirectory with custom ignores
codebase-context /path/to/project output.txt src "*.test.js,*.spec.js,__tests__"

Configuration

The tool comes with sensible defaults that work for most projects:

Default Limits

• Maximum file size: 1MB
• Maximum line count: 5000 lines

Default Ignore Patterns

The tool automatically ignores common patterns including:

• Package management: node_modules, package-lock.json, etc.
• Version control: .git, .gitignore
• Build outputs: dist, build, .next, etc.
• System files: .DS_Store, Thumbs.db
• Environment/secrets: .env files, .pem, .key
• Logs and temp files: logs, *.log, temp, tmp
• Language-specific: __pycache__, *.class, etc.
• Binary files: Images, videos, archives, executables

Output Format

The tool produces an XML-formatted file with each source file enclosed in <file> tags:

<file path="src/index.js">
// File content goes here
</file>

<file path="src/utils/helpers.js">
// Another file's content
</file>

This format is optimized for use with LLMs as it clearly delineates file boundaries and preserves path information.

Use Cases

• Get help with bugs by providing full context to an LLM
• Ask for architecture recommendations with full codebase knowledge
• Generate documentation based on your entire project
• Request refactoring suggestions with proper context

Contributing

Contributions are welcome! Feel free to submit issues or pull requests.

License

ISC