InferAct: Inferring Safe Actions for LLM-Based Agents Through Preemptive Evaluation and Human Feedback

Open in new window