repomix

Rather than manually copying individual files into a chat interface, Repomix traverses the repository, filters out irrelevant files (binaries, build artifacts, dependencies), and produces a structured output with file paths preserved as headers allowing the AI to reason about the entire codebase with its original file organization intact.

The tool is configurable to respect .gitignore rules, exclude specific directories, and apply custom filters so the output focuses on the code that matters.

Repomix includes line count statistics and token estimates for the packed output, helping users understand how much context the resulting file will consume when submitted to different models.

Output formats include plain text, XML (optimized for Claude's XML-aware context handling), and Markdown, each structured to maximize how effectively the AI can reference specific sections.

Repomix is open-source and available as an npm package, installable globally with a single command.

It is particularly useful for tasks like asking an AI to explain an unfamiliar codebase, generating comprehensive documentation, identifying architectural inconsistencies across files, or performing security audits across the full project.

As LLM context windows continue to expand, tools like Repomix that efficiently structure repository content for AI consumption are becoming a standard part of AI-augmented development workflows.