Search for a command to run...
ToolHazard: Scaling Adversarial Environments for Security Evaluation and Alignment of LLM-based Agents