Repomix
github.com/yamadashy/repomix- Category
- Developer Tools
- Rank
- No. 75Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- yamadashy
- GitHub
- 28.3k stars
- Latest release
- v1.18.0
- Date
About
A powerful tool that packs your entire repository into a single, AI-friendly file format. Perfect for feeding your codebase to Large Language Models like Claude, ChatGPT, or other AI tools for code analysis and refactoring.
What it does
Repomix gathers selected files from local or remote repositories, applies ignore and credential checks, then produces XML, Markdown, JSON, or plain text output. It also reports token counts and can extract key code elements with Tree-sitter compression.
Why it's ranked here
The tool covers the practical details around preparing repository context: selection, exclusions, token measurement, secret detection, remote sources, and multiple output formats. Its command-line, library, website, and browser entry points make the same core job useful across automated and interactive workflows.
What's good
Git-aware filtering respects standard ignore rules and a dedicated Repomix ignore list. Secretlint excludes files matching known credential formats. Per-file and repository token counts make context budgeting visible, while include patterns, ignore patterns, piped file lists, and remote revision selection give precise control over scope.
Tradeoffs
The command-line package requires Node 22 or newer. Secret detection targets known credential formats, so it should not replace deliberate review before sharing output. Compression extracts key elements rather than retaining every source detail. Token counting measures context size but does not make an oversized repository fit automatically.
How to use it well
Use it when an engineer or automation needs a controlled repository snapshot for an AI assistant. Start with ignore rules, narrow large projects through patterns or piped file lists, inspect the security findings, and use token counts to tune scope. It prepares context; it does not perform the subsequent analysis, refactoring, or code review itself.
Technical notes+
package.json defines a public TypeScript-built package with JavaScript and declaration outputs, command-line packaging, Node 22 minimum support, and MIT licensing. src/index.ts exposes packing, file collection and processing, remote parsing, security scanning, token counting, Tree-sitter parsing, configuration, CLI actions, and a unified worker handler as library exports. vitest.config.ts targets tests/**/*.test.ts and collects coverage from src/**/* except src/index.ts. browser/wxt.config.ts defines Chrome and Firefox extension metadata with GitHub-only host access. website/server/src/index.ts builds a Hono server with CORS, compression, request timeout, bot protection, rate limiting, Turnstile, a health route, and the POST /api/pack endpoint.
Observed
- License
- MIT
- Primary language
- TypeScript
- Packaging
- Public npm package with JavaScript output, TypeScript declarations, and a command-line executable; npm, Yarn, Bun, Homebrew, and Docker usage are documented.
- Interfaces
- Command-line tool, importable library, website, HTTP packing endpoint, and Chrome and Firefox browser extensions.
- Runtime
- The package requires Node.js 22 or newer.
- Testing
- Vitest is configured for Node tests under the tests directory, with source coverage reporting.
Read from README.md, package.json, src/index.ts, vitest.config.ts, browser/types.d.ts, browser/wxt.config.ts, browser/vitest.config.ts, website/server/src/index.ts, LICENSE, AGENTS.md, CLAUDE.md, flake.nix, Dockerfile, biome.json, flake.lock.
What it can do
Convert entire repository into single AI-friendly file
Code repository → Single packaged file optimized for AI consumption
Count tokens in codebase
Repository files → Token count statistics
Filter files using git awareness
Repository with git configuration → Filtered file collection respecting gitignore rules
Perform security checks on code
Source code files → Security analysis results
Compress code using Tree-sitter parsing
Source code files → Compressed code representation
Prepare codebase for Large Language Model analysis
Software repository → LLM-ready formatted codebase
Tags
Tech Stack
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.