This tool transforms local codebases into Markdown files, ideal for feeding into Large Language Models (LLMs) for analysis, summarization, or code generation tasks. It's designed for developers and researchers working with AI code assistants or needing to quickly represent project structures for LLM consumption.

How It Works

Codefetch recursively scans a specified directory, respecting .gitignore and .codefetchignore files. It then converts the collected code files into a single Markdown document, optionally including line numbers and a project directory tree. Users can filter by file extensions, include/exclude specific files or directories, and select tokenization methods compatible with various LLM models.

Quick Start & Requirements

• Install and run via npx codefetch.
• Requires Node.js.
• Official docs: https://github.com/regenrek/codefetch#readme

Highlighted Details

• Supports multiple token encoders (simple, p50k, o200k, cl100k) for LLM compatibility.
• Offers token limiting strategies (sequential, truncated) to manage output size.
• Allows custom prompts and integrates with AI tools like Cursor.
• Can generate a visual project tree structure within the Markdown output.

Maintenance & Community

• Developed by regenrek.
• Links to X/Twitter (@kregenrek) and Bluesky (@kevinkern.dev) provided.
• Inspired by codetie and sitefetch CLIs.

Licensing & Compatibility

• The README does not explicitly state a license.

Limitations & Caveats

The project's license is not specified, which may impact commercial use or integration into closed-source projects.