ИИ-агенты
garak
Scan your LLM for vulnerabilities before attackers do.
★ 9,5 тыс. звёздApache-2.0
Смотреть руководство →modelscan — проект с открытым исходным кодом из категории «ИИ-агенты» от protectai (protectai/modelscan на GitHub). У него 781 звезда на GitHub и 168 форков, он распространяется по лицензии Apache-2.0 и последний раз обновлялся на GitHub 28 сентября 2026.
Scan AI models for serialization attacks before you load them.
Decide what you want the project to do before installing anything.
Grab the source and skim the README for setup requirements.
Install dependencies and add any API keys or settings it needs.
Launch it and follow the example workflow from the docs.
Inspects model files for dangerous payloads.
Scan before deserialization, not after.
Trust-but-verify for downloaded weights.
Gate deployments on scan results.
Drops into ML pipelines easily.
From a dedicated AI-security company.
Download the source and move into the project folder.
git clone https://github.com/protectai/modelscan.git
cd modelscanIsolate dependencies, then install requirements.
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txtLaunch using the entry point shown in the README.
python -m modelscan # check README for the exact commandIt is an open-source project at protectai/modelscan with 168 forks. Check the contributors tab and recent commits to judge how active maintenance is.
Yes — Apache 2.0 allows commercial use and modification. Keep the license and notice files, and note it includes a patent grant.
Model serialization attacks — malicious code hidden in model files that executes on load.
Point it at model files before loading them, ideally as a CI step.
Clone the repo and follow the README quickstart — the docs at http://modelscan.ai walk through setup.
Scan your LLM for vulnerabilities before attackers do.
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…
Connect your browser to AI models with a Dia-style assistant.
Test your agents against prompt-injection attacks.
A coding agent tuned for open models like Kimi, GLM, DeepSeek, and Qwen.
Hugging Face's LeRobot: datasets, models and tools to make real-world robotics AI accessible
Другие проекты из категории «ИИ-агенты» с наибольшим числом звёзд в каталоге LibreAI — реалистичный список альтернатив, если modelscan вам не подходит.
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
Соседние проекты из категории «ИИ-агенты» в каталоге, ближайшие по числу звёзд к modelscan — продолжайте изучать полку по обе стороны от него.
Open-source conversational voice AI agents.
SWE-Gym: the first training environment for real-world software engineering agents and verifiers (ICML 2025).
An MCP server that puts Hyperbrowser's browsers behind the protocol.
See what your AI agents do — at the system level.
SimWorld: an Unreal Engine 5 simulator for evaluating LLM/VLM agents in physical and social worlds.
A very large curated list of AI tools, updated for 2026.
Бесплатно и открыто — забирайте исходный код, ставьте звезду репозиторию и начинайте создавать.