DEV.co
AI Frameworks · yamadashy

repomix

Repomix is a CLI and web tool that packages your entire codebase into a single AI-friendly file (XML, Markdown, or plain text) for feeding to LLMs like Claude or ChatGPT. It includes token counting, gitignore awareness, optional code compression via Tree-sitter, and built-in secret detection using Secretlint.

Source: GitHub — github.com/yamadashy/repomix
26.9k
GitHub stars
1.4k
Forks
TypeScript
Primary language
MIT
License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

FieldValue
Repositoryyamadashy/repomix
Owneryamadashy
Primary languageTypeScript
LicenseMIT — OSI-approved
Stars26.9k
Forks1.4k
Open issues168
Latest releasev1.16.0 (2026-06-29)
Last updated2026-07-05
Sourcehttps://github.com/yamadashy/repomix

What repomix is

TypeScript-based Node.js tool that traverses repositories, respects .gitignore/.repomixignore rules, optionally compresses code structures via Tree-sitter AST extraction, counts tokens per file, and integrates Secretlint for credential/secret detection before packaging into templated output formats.

Quickstart

Get the repomix source

Clone the repository and explore it locally.

terminalbash
git clone https://github.com/yamadashy/repomix.gitcd repomix# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Codebase context for LLM-assisted refactoring

Developers can pipe the generated file into Claude, ChatGPT, or other LLMs with natural-language refactoring requests, allowing the AI to understand full code structure and generate targeted changes across multiple interdependent files.

Rapid LLM-powered code review and analysis

Engineering teams can share entire repositories with AI tools for comprehensive code reviews, architectural analysis, or vulnerability scanning without manual file selection or context loss.

Token-efficient repository exploration

The token counting and optional compression features help developers optimize LLM context windows, making large codebases analyzable within model limits by selectively reducing boilerplate while preserving logic.

Implementation considerations

  • Requires Node.js runtime; available via npm, yarn, bun, or Homebrew. npx repomix@latest allows zero-install usage.
  • Configure .repomixignore rules early to exclude node_modules, build artifacts, vendor deps, and sensitive files; default .gitignore respect is insufficient for many workflows.
  • Token counting is an estimate; actual usage varies by LLM tokenizer (GPT vs. Claude vs. others). Test on real models before relying on counts for context budgeting.
  • Secretlint is built-in but rule coverage is finite; manual review of sensitive repos is strongly recommended before sharing output with external AI services.
  • Code compression via Tree-sitter requires language support; unsupported or dynamic languages may fail compression silently or partially.

When to avoid it — and what to weigh

  • Extremely large monorepos without careful filtering — Packaging a 1GB+ monorepo with full history and dependencies without .repomixignore rules can create output files too large for most LLM token limits. Requires disciplined configuration to be practical.
  • Sensitive production data or credentials in codebase — While Secretlint integration mitigates risk, it is not foolproof. Code containing hardcoded secrets, API keys, or proprietary algorithms should be reviewed before sending output to external LLM services.
  • Need for incremental or delta-based updates — Repomix re-packages the entire repository each time; it does not support delta-only or change-based output, making it less suitable for continuous or real-time code analysis workflows.
  • Binary or non-text file-heavy repositories — Repomix is designed for source code; repositories with heavy binary assets, images, or compiled artifacts will bloat output without meaningful AI value.

License & commercial use

MIT License. Permissive OSI-approved license allowing unrestricted use, modification, and distribution for commercial and private purposes with minimal obligations (attribution and license notice preservation).

MIT license permits commercial use without restriction. No license-based barriers to integration into proprietary products or services. However, verify third-party dependencies (Secretlint, Tree-sitter, etc.) for their own license terms to ensure full compliance in commercial contexts.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

SignalAssessment
MaintenanceActive
DocumentationAdequate
License clarityClear
Deployment complexityLow
DEV.co fitGood
Assessment confidenceHigh
Security considerations

Secretlint integration mitigates accidental credential leakage but is not a guarantee. Tool processes local filesystem and outputs to file or external web service (repomix.com). If sharing output with LLM APIs (OpenAI, Anthropic, etc.), assume standard SaaS data residency and retention policies apply. Audit .repomixignore rules to exclude sensitive paths. Binary executable or package dependencies (Tree-sitter, Secretlint) should be verified from trusted sources.

Alternatives to consider

LLM-specific context tools (Claude prompt caching, GPT system messages)

OpenAI and Anthropic offer native context management features. If locked to one model, built-in tools may reduce friction vs. generic Repomix output. However, Repomix is model-agnostic and format-flexible.

Git-based log analysis (e.g., git-sitter, conventional-changelog)

Focused on commit history and change tracking. Better for understanding evolution; weaker for structural code review. Complementary rather than competitive.

Custom AST extraction or code indexing (Tabnine, Sourcegraph)

More sophisticated semantic analysis and caching. Typically closed-source or cloud-hosted. Overkill for simple packing; stronger for IDE integration and real-time suggestion.

Software development agency

Build on repomix with DEV.co software developers

Try Repomix today via npx repomix@latest or visit repomix.com. Configure .repomixignore for your repository, generate your first output, and feed it to your favorite LLM.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

Related on DEV.co

Explore the category and the services that help you build with it.

repomix FAQ

Does Repomix send my code to external servers?
The CLI tool processes code locally and outputs to a file by default. The web UI (repomix.com) processes repos via the website servers; privacy policy should be reviewed if using it with proprietary code. Browser extension and VSCode extension vary; check their documentation.
What output formats are supported?
XML (default), Markdown, and plain text. Web UI allows format selection. CLI outputs XML by default; other formats require configuration or custom post-processing.
How accurate is the token count?
Token counts are estimates based on a generic tokenizer or per-file heuristics. Actual usage depends on the target LLM's tokenizer (GPT, Claude, Llama, etc.). Always validate on the real model before relying on counts for context budgeting.
Can Repomix be used in CI/CD pipelines?
Yes. The CLI is scriptable and outputs to files. No interactive prompts or network dependencies required (unless using remote Secretlint rules). Suitable for automated codebase analysis, archival, or passing to downstream LLM analysis jobs.

Work with a software development agency

Need help beyond evaluating repomix? DEV.co is a software development agency offering software development services and web development for teams of every size. Our software developers and web developers build custom software, web applications, APIs, and ai frameworks integrations — and maintain them long-term.

Ready to streamline LLM-assisted code review?

Try Repomix today via npx repomix@latest or visit repomix.com. Configure .repomixignore for your repository, generate your first output, and feed it to your favorite LLM.