jason-oneal/secgpt-orchestrator

该数据集包含用于授权渗透测试和安全评估工作流的结构化编排计划。其设计目的是训练和评估大型语言模型，使其能够作为规划者和协调者，而非漏洞利用执行者。数据集强调决策制定、工具排序、范围处理和安全终止行为。数据集明确避免了有效载荷、漏洞利用或漏洞武器化。数据格式为JSONL，每行包含一个JSON对象，描述了评估目标和计划步骤。

This dataset contains structured orchestration plans for authorized penetration testing and security assessment workflows. It is designed to train and evaluate large language models that act as planners and coordinators, not exploit executors. The focus is on decision-making, tool sequencing, scope handling, and safe termination behavior. This dataset explicitly avoids payloads, exploits, or vulnerability weaponization. Data format is JSONL, each line contains a single JSON object describing the assessment objective and plan steps.