PostExploitBench is a gated cybersecurity benchmark dataset for controlled multi-host post-exploitation evaluation. It contains isolated range artifacts, challenge metadata, vulnerable service files, verification assets, and supporting files for reproducible research and safety evaluation.
The dataset is evaluation-only. It must not be used for model training, fine-tuning, reinforcement learning, dataset construction, retrieval corpus construction, agent… See the full description on the dataset page:
https://huggingface.co/datasets/AgentCyberRange/PostExploitBench.
