---
格式版本: 2
标题: "Demo: tfdrift - A Severity Taxonomy and Risk Classification Framework for Infrastructure Drift Detection"
原文链接: "https://arxiv.org/abs/2608.18173"
发布日期: "2026-08-17"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Mon, 17 Aug 2026 19:12:06 UTC (5 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T14:41:14+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T14:38:30+08:00"
入库时间: "2026-08-20T06:41:14.295Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=AWS&searchtype=all"
匹配关键词:
  []
相关厂家:
  - "AWS"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 0
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "论文主题为基础设施即代码配置漂移检测，属软件工程领域，与超节点/AI Rack/机柜级AI基础设施完全无关，无任何相关技术、商业或部署信息。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T14:42:38+08:00"
AI主题相关性: 0
AI来源权威性: 0
AI新颖性: 0
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 0
AI摘要: "针对Terraform基础设施漂移检测中的告警疲劳，作者提出tfdrift开源框架，按资源类型和属性影响对漂移变更做严重性风险分级。在AWS工作区评估中，严重性过滤使告警量减少73%，同时保留94%的安全相关变更。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:25:12.532Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.18173"
---

## Computer Science > Software Engineering

## Title:Demo: tfdrift - A Severity Taxonomy and Risk Classification Framework for Infrastructure Drift Detection

Authors:[Sudarshan Bhagvanthakur](https://arxiv.org/search/cs?searchtype=author&query=Bhagvanthakur,+S)

[View PDF](https://arxiv.org/pdf/2608.18173) [HTML (experimental)](https://arxiv.org/html/2608.18173v1)

> Abstract:Infrastructure as Code (IaC) tools like Terraform have become the standard for declarative cloud resource management, yet configuration drift, where deployed infrastructure diverges from its declared state, remains a persistent operational and security challenge. Current detection approaches treat all changes equivalently, contributing to alert fatigue that causes operators to miss security-critical modifications. We propose a generalized severity taxonomy for infrastructure drift that classifies changes into four risk tiers based on resource type and attribute-level impact. We implement this taxonomy in tfdrift, an open-source classification framework with 60+ configurable rules covering AWS, Azure, and GCP resource patterns (evaluation reported here is AWS-focused). Evaluation across 150+ AWS Terraform workspaces demonstrates that severity filtering reduces alert volume by 73% while retaining 94% of security-relevant changes, offering a lightweight alternative to ML-based alert filtering. tfdrift is available at [this http URL](http://github.com/sudarshan8417/tfdrift).

| Comments: |  |
| --- | --- |
| Subjects: | Software Engineering (cs.SE) |
| Cite as: | [arXiv:2608.18173](https://arxiv.org/abs/2608.18173) \[cs.SE\] |
|  | (or [arXiv:2608.18173v1](https://arxiv.org/abs/2608.18173v1) \[cs.SE\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.18173](https://doi.org/10.48550/arXiv.2608.18173) |

## Submission history

From: Sudarshan Bhagvan Thakur \[[view email](https://arxiv.org/show-email/f2c52d1e/2608.18173)\]  
**\[v1\]** Mon, 17 Aug 2026 19:12:06 UTC (5 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.18173) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))