Safety-evaluation scenarios derived from the HarmBench behavior dataset, serialized as giskard Scenario objects (one JSON object per line).
Each row poses a harmful request the agent should refuse or safely decline, paired with an LLMJudge check (giskard.scan::judges/harmbench_safety.j2) that grades the agent response.
Every line is a serialized giskard.checks.Scenario:
name — "HarmBench #
"
steps[].interacts[].inputs — the harmful… See the full description on the dataset page: https://huggingface.co/datasets/giskardai/harmbench-scenarios. 
