// agents · NVIDIA/garak

garak

garak é um projeto de agentes de IA de código aberto de NVIDIA (NVIDIA/garak no GitHub). Tem 9.497 estrelas no GitHub e 1.349 forks, é publicado sob a licença Apache-2.0 e foi atualizado pela última vez no GitHub em 7 de outubro de 2026.

Scan your LLM for vulnerabilities before attackers do.

★ 9,5 mil estrelas⑂ 1,3 mil forksApache-2.0atualizado out 2026
▦ visual: o README deste projeto não inclui nenhuma captura incorporável,
então aqui vai um marcador. O guia abaixo está completo mesmo assim.
// dados do projeto

Dados do projeto

Repositório
NVIDIA/garak
Estrelas no GitHub
9.497
Forks
1.349
Licença
Apache-2.0
Última atualização no GitHub
7 de outubro de 2026
Categoria
Agentes de IA
Tipo de ficha
Guia de instalação completo
// como funciona

Como funciona

  1. Define what you want

    Write down the job you will use the project for before touching any code.

  2. Clone the repository

    Download the source and read the README's setup section first.

  3. Install dependencies

    Run the install steps and add any required API keys or config.

  4. Run the quickstart

    Follow the example in the docs to confirm everything works.

// recursos

Por que as pessoas gostam tanto

LLM vulnerability scanning

Probes models for known weakness classes.

NVIDIA-backed

Maintained under NVIDIA's organization.

Security focus

Built for pre-deployment adversarial testing.

Python tooling

Scriptable and extensible in Python.

Apache-2.0

Permissive licensing.

Active community

Discord plus 1,300+ forks.

Advertisement
// execute você mesmo

Execute você mesmo

  1. Step 1 — Install

    Install garak from PyPI.

    pip install -U garak
  2. Step 2 — Pick a target

    Choose the model or endpoint you want to probe.

    # local model or API endpoint — see docs
  3. Step 3 — Run a scan

    Launch a probe run per the documentation.

    garak --help
    # see docs for scan commands
// casos de uso

O que você pode criar

  • Red-team a chatbot before launch.
  • Test a model for jailbreak susceptibility.
  • Audit third-party LLM integrations.
  • Add security scans to your LLM release process.
Advertisement
// perguntas frequentes

Perguntas frequentes

Who maintains garak?

It is maintained under NVIDIA at NVIDIA/garak, with 1,349 forks.

Is garak free to use?

Yes. It is released under the Apache-2.0 license, which permits commercial use, modification, and redistribution. Read the LICENSE file for the exact terms.

Do I need a paid API to use it?

Scanning itself is local; you need access to the target model, via API or a local deployment.

How does it compare to other red-teaming tools?

Compare on probe coverage and reporting for the model types you ship.

How do I get started?

Install the package and follow the docs to run your first scan against a test model.

// continue explorando

Ferramentas semelhantes

Agentes de IA

modelscan

Scan AI models for serialization attacks before you load them.

★ 781 estrelasApache-2.0
Ver o guia →
Agentes de IA

vigil-llm

Detect prompt injections and jailbreaks before they hit your LLM.

★ 504 estrelasApache-2.0
Ver o guia →
Agentes de IA

career-ops

Open-source AI job search agent and job finder: scan job boards, score each job 1-5 against your CV before you apply, tailor an ATS-friendly resume and cover…

★ 73,4 mil estrelasMIT
Ver o guia →
Agentes de IA

macos-harness

A minimal harness that lets an LLM fully control a Mac.

★ 902 estrelasMIT
Ver o guia →
Agentes de IA

deepeval

Unit tests for LLMs — evaluate before you ship.

★ 18,7 mil estrelasApache-2.0
Ver o guia →
Agentes de IA

browser-harness

A self-healing harness that connects an LLM directly to your real browser

★ 18,4 mil estrelasMIT
Ver o guia →
// alternativas

Alternativas em Agentes de IA

Os outros projetos de Agentes de IA com mais estrelas no catálogo do LibreAI — a lista realista de alternativas se garak não for o ideal para você.

  • openclaw★ 391,6 mil

    A personal AI that really does things — on any OS, with your data staying yours.

  • superpowers★ 294,9 mil

    An agentic skills framework & software development methodology that works.

  • ECC★ 275,2 mil

    Agent harness optimization system with skills, memory and security for Claude Code, Codex and more

  • hermes-agent★ 252,0 mil

    The agent that grows with you — a Python framework for adaptive AI agents.

  • deepseek-harness★ 245,5 mil

    DeepSeek's plugin-based agent harness where everything is a plugin.

// mais em agents

Mais em Agentes de IA

Projetos de Agentes de IA próximos no catálogo, os mais próximos em estrelas de garak: continue explorando dos dois lados dele.

  • nasiko★ 9,5 mil

    The open runtime for AI agents — a Rust platform for running, governing and routing agents with TokenOps.

  • MobileAgent★ 9,3 mil

    Mobile-Agent — multimodal GUI agents that operate phones and PCs from screenshots and instructions.

  • mobilerun★ 9,6 mil

    Control Android and iOS devices with natural language — an LLM-agnostic mobile agent framework and CLI.

  • local-deep-research★ 9,2 mil

    Deep research that stays on your machine — local or cloud LLMs, 10+ search engines, encrypted and private.

  • UFO★ 10,0 mil

    Microsoft's GUI agent that grew into UFO³ — orchestrating agents across Windows, Linux and mobile devices.

  • evolver★ 9,1 mil

    A GEP-powered self-evolving engine for AI agents — auditable evolution with Genes, Capsules and Events.

Baixe o garak

Grátis e aberto: pegue o código-fonte, dê uma estrela no repositório e comece a criar.