v1.0.1

v1.0.0

• Initial release
• 80+ adversarial attack implementations across 10 categories
• Unified interface for various LLM providers
• Safety assessment tools for detecting harmful outputs
• Pipeline orchestration for attack generation, testing, and evaluation
• Support for multiple LLM providers (OpenAI, Anthropic, Google, Yandex, GigaChat, Ollama)
• Comprehensive documentation and examples

v1.0.1

• Updated quickstart documentation
• Added ability to use Ollama and other models without adding model name to constants
• Fixed AIM attack implementation