galcan/terraform_sec

Terraform安全数据集是一个包含62,406个Terraform项目的综合数据集，使用tfsec进行安全漏洞分析。该数据集旨在训练大型语言模型（LLMs）理解、识别和修复Terraform基础设施即代码中的安全问题。数据集包含44,030个安全项目（70.6%）和18,376个不安全项目（29.4%），格式为JSONL（JSON Lines）。每个示例包含项目ID、指令、输入（完整的Terraform配置）、输出（安全分析结果）和元数据（如文件数量、问题数量、是否安全等）。数据集覆盖了AWS、Azure、GCP和Kubernetes等多个云提供商和服务的安全漏洞，包括加密问题、访问控制、日志和监控、网络安全和最佳实践等常见漏洞。

The Terraform Security Dataset is a comprehensive dataset of 62,406 Terraform projects analyzed for security vulnerabilities using tfsec. This dataset is designed for training Large Language Models (LLMs) to understand, identify, and fix security issues in Terraform infrastructure-as-code. It includes 44,030 secure projects (70.6%) and 18,376 insecure projects (29.4%), formatted as JSONL (JSON Lines). Each example contains a project ID, instruction, input (complete Terraform configuration), output (security analysis result), and metadata (e.g., file count, issue count, is_secure flag). The dataset covers security vulnerabilities across multiple cloud providers and services, including AWS, Azure, GCP, and Kubernetes, addressing common issues such as encryption problems, access control, logging and monitoring, network security, and best practices.