repomix

Repomix packages entire GitHub repositories into single AI-friendly files, preserving directory structure and metadata. It converts codebases into formats like XML, Markdown, JSON, or plain text optimized for LLM context loading, enabling AI agents to analyze large projects efficiently without overwhelming token limits.

Repository packing outputs multiple formats: XML, Markdown, JSON, and plain text. Each format preserves file structure and context while optimizing for different use cases—XML for structured parsing, Markdown for readability, JSON for programmatic access, and plain text for minimal overhead.

Yes, batch processing automates packing of multiple repositories simultaneously, ideal for large-scale code analysis, building AI training datasets, or conducting systematic security audits across many codebases without manual repetition.

Built-in token counting tracks the size of packaged repositories relative to AI model context limits. This ensures packed codebases fit within LLM token budgets before analysis, preventing context overflow and enabling efficient context allocation for large projects.

Repository packing supports both remote GitHub repositories and local codebases. This flexibility lets you package public projects, private repositories, or on-premise code without requiring repository hosting or external access for every analysis.

Built-in security checks scan packaged repositories for common vulnerabilities and sensitive data patterns during the packing process. This protects against accidentally exposing credentials, API keys, or security misconfigurations when preparing codebases for AI analysis or sharing.