Why is token counting important in Repomix?

LLMs have limited context windows (e.g., 200k for Claude). Repomix counts tokens so you can ensure your packaged codebase fits within the limits of your target model.

What is Repomix used for?

Repomix packs an entire repository into a single, structured file that is easy for AI models to understand, making it ideal for codebase analysis, debugging, and providing context to LLMs.

Can I use Repomix on remote repositories?

Absolutely. You can use the --remote flag to package any public GitHub repository without needing to clone it to your local machine first.

Does Repomix protect my sensitive data?

Yes, it includes built-in security checks using Secretlint to detect and warn you about API keys, passwords, or credentials before you package your code for AI analysis.

Which output formats does Repomix support?

It supports XML, Markdown, JSON, and plain text, allowing you to choose the format best suited for your specific AI assistant or developer workflow.

Repomix Repository Packager

Name: Repomix Repository Packager
Author: nordeim

bynordeim

0•

开发者工具

Packages entire codebases into optimized, AI-friendly files to provide rich context for LLMs like Claude, ChatGPT, and Gemini.

Repomix is a specialized tool designed to bridge the gap between complex codebases and AI context windows. It intelligently packs entire repositories—including local or remote sources—into single, structured files such as XML, Markdown, or JSON while respecting .gitignore rules and security constraints. By optimizing file structures, stripping unnecessary comments, and providing accurate token counts, it allows developers to feed large-scale code context into LLMs for tasks like architectural analysis, security auditing, and comprehensive bug investigation without the need for manual copy-pasting.

主要功能

01Remote repository support to process GitHub repos without local cloning.

020 GitHub stars

03Multi-format output including XML, Markdown, and JSON optimized for LLM consumption.

04Intelligent token counting to manage LLM context window limits.

05Built-in security scanning with Secretlint to prevent sensitive data leakage.

06Customizable file filtering and comment removal to reduce noise and token usage.

使用场景

01Generating comprehensive context for AI-driven code reviews and architectural planning.

02Creating repository snapshots for documentation and codebase investigation.

03Packaging third-party libraries for security audits and vulnerability assessments.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nordeim/prompt-engineering repomix

For use in Claude.ai and ChatGPT

Download Skill