repomix_skill

This skill packages entire repositories into AI-friendly single files, generating context, token counts, and multi-format outputs for efficient LLM analysis.

Python

12

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill samhvw8/dotfiles --skill repomix

SKILL.md10.2 KB

Overview

This skill packages repositories into single, AI-friendly files to feed Large Language Models and other analysis tools. It produces multiple output formats, counts tokens, and prepares codebase snapshots for code review, security audits, or implementation planning. The tool supports local and remote repos, configurable include/exclude patterns, and optional comment stripping for compact context.

How this skill works

Repomix scans the target repository (local or remote), applies include and ignore rules, and formats files into a single output with clear separators and optional line numbers. It computes token counts per file and for the whole package to respect LLM context windows and can run security checks to flag secrets. Outputs can be exported as XML, Markdown, JSON, or plain text and copied to the clipboard or written to a specified path.

When to use it

Preparing repository context for LLM consumption (ChatGPT, Claude, Gemini)
Packaging codebases for security audits or third-party library review
Generating repeatable codebase snapshots for documentation or release notes
Feeding focused modules to AI for bug investigation or implementation planning
Reducing payload size by removing comments or excluding large artifacts

Best practices

Start with the default run and refine include/exclude patterns iteratively
Maintain a .repomixignore for sensitive files and large artifacts
Enable security checks when analyzing unfamiliar or third-party code
Monitor token counts and choose output style to match the target LLM
Exclude node_modules, build artifacts, and media files to keep output concise

Example use cases

Package a feature branch into markdown for an AI-driven code review
Generate an XML snapshot of a vendor library for a security audit
Create a focused context of src/auth and src/api directories to troubleshoot a bug
Produce a JSON bundle for programmatic analysis in CI pipelines
Remove comments and copy the packaged output to the clipboard for quick paste into Claude

FAQ

Yes — provide a shorthand owner/repo, full URL, or a commit URL to package a remote repo directly.

How does token counting help with LLMs?

Token counts per file and total help you stay within model context limits and choose which files to include or split.

What output format should I choose for AI vs human readers?

Use XML for structured AI consumption, JSON for programmatic workflows, and Markdown for readable, syntax-highlighted context.