Tag: red-teaming

Discover Giskard v3, the modular async-first open-source library for testing, evaluating, and red teaming LLM agents. Stop shipping broken AI with systematic scenario-based evaluation.

Cochise is a 576-line Python prototype that uses LLMs to autonomously compromise Active Directory networks. Full domain admin in 90 minutes for $2. Published in ACM TOSEM with reproducible benchmarks across Claude, GPT, Gemini, and DeepSeek.