Trustworthy Agentic RAG for Enterprise AI: A Systematic Survey and Framework for Security, Privacy, Robustness, Governance, Evaluation, and Cloud-Native Deployment — Companion Repository

Companion artifacts for the survey article Trustworthy Agentic RAG for Enterprise AI: A Systematic Survey and Framework for Security, Privacy, Robustness, Governance, Evaluation, and Cloud-Native Deployment, submitted to MDPI Electronics (Special Issue: Trustworthy Retrieval-Augmented Generation — Security, Privacy, Robustness, and Fairness in LLM Systems). The deposit releases the systematic-review protocol, search strings, screening log, BibTeX bibliography, and reusable artifacts of the TERA (Trustworthy Enterprise RAG Agents) framework: an eight-dimension taxonomy, a threat-to-control matrix, an agentic-RAG evaluation matrix, a cloud-native reference architecture component model, a thirty-two-control governance and human-oversight checklist, and a research roadmap.本数据集为投稿至MDPI《电子学》特刊（特刊主题：大语言模型（LLM）系统中的可信检索增强生成——安全性、隐私性、鲁棒性与公平性）的综述文章《面向企业人工智能的可信智能体检索增强生成（Trustworthy Agentic RAG）：面向安全性、隐私性、鲁棒性、治理、评估与云原生部署的系统综述及框架》的配套附属资料。 本次存档公开了TERA（可信企业检索增强生成智能体，Trustworthy Enterprise RAG Agents）框架的系统综述方案、检索式、筛选日志、BibTeX参考文献库及可复用附属资料，该框架涵盖八维分类法、威胁-管控矩阵、智能体检索增强生成评估矩阵、云原生参考架构组件模型、含32项管控措施的治理与人工监督核查清单以及研究路线图。