A lightweight Windows tool for scanning and classifying sensitive data in files. Single executable, zero installation. Automatically detects personal information, credentials, and secrets, with a built-in web UI and report export.
轻量级 Windows 敏感数据扫描与分类分级工具。单文件运行，无需安装，自动识别文件中的个人信息、密钥凭证等敏感数据，提供可视化界面与报告导出。
| 10 Built-in Detection Rules / 10 种内置检测规则 | ID cards, phone numbers, bank cards, emails, IPs, USCC, AWS keys, GitHub tokens, private keys, high-entropy secrets |
| 5-Level Classification / 5 级分类标准 | L1 Public → L2 Internal → L3 Confidential → L4 Secret → L5 Restricted |
| 19 File Formats / 19 种文件格式 | Plain text (.txt/.csv/.log/.json/.xml/.md/.env/.yaml/.yml/.ini/.conf/.toml/.bat/.ps1/.sh/.sql) + Office (.docx/.xlsx/.pptx) |
| Multi-core Parallel Scanning / 多核并行扫描 | Auto-utilizes multi-core CPU with real-time progress updates |
| False Positive Correction / 误报修正 | Single or batch correction of classification results; exports include both original and corrected results |
| Visual Statistics / 可视化统计 | Donut chart for level distribution, bar chart for category breakdown, summary cards |
| Report Export / 报告导出 | JSON (for programmatic use) / CSV (with UTF-8 BOM for Excel) |
| Fully Local / 纯本地运行 | All scanning and analysis runs locally — no internet, no data upload |
| Single Executable / 单文件交付 | Compiles to ~7 MB .exe, double-click to run, auto-opens browser UI |
- Download the latest
datasentinel.exefrom Releases - Double-click to run — the browser opens automatically
从 Releases 下载最新的 datasentinel.exe，双击运行，浏览器自动打开操作界面。
If the browser doesn't open automatically, check
datasentinel.login the same directory for the access URL.如果浏览器未自动打开，查看程序同目录下
datasentinel.log获取访问地址。
- Select target directory / 选择目标目录 — Enter a path or browse for the folder to scan
- Configure scan parameters / 配置扫描参数 — Choose file types and detection rules
- Start scan / 开始扫描 — Wait for progress to complete
- View results / 查看结果 — Filter by level, sort, click "View" for match details
- Correct false positives / 修正误报 — Single or batch correction of classification levels
- Export report / 导出报告 — Download JSON or CSV report
Requirement / 环境要求: Go 1.21+
# Standard build (with console window for debugging)
# 标准编译（带控制台窗口，便于调试）
go build -o datasentinel.exe .
# Release build (no console window, for distribution)
# 发布编译（无控制台窗口，适合分发）
go build -ldflags="-s -w -H windowsgui" -o datasentinel.exe .WalkFiles (goroutine) ──→ Worker Pool (N=NumCPU) ──→ Result Channel ──→ Report
Recursive walk ExtractText() Collect+Summarize SSE Push
Extension filter Rule.Match() computeSummary
Classify() → FileResult
- Go Backend — HTTP server + goroutine concurrent pipeline + context cancellation
- System Browser UI — Local HTTP server with automatic browser launch
- SSE Real-time Push — Server-Sent Events for scan progress
- embed.FS — HTML/CSS/JS embedded in binary for single-file delivery
- Standard Library Only — Office document parsing via
archive/zip+encoding/xml, no heavy dependencies
├── main.go # Entry point: HTTP server + browser launch / 入口：启动 HTTP 服务 + 打开浏览器
├── model/types.go # Data structures: Level, Match, FileResult, ScanReport / 数据结构
├── rules/
│ ├── rules.go # Rule definitions, regex matching, masking / 规则定义、正则匹配、脱敏处理
│ ├── patterns.go # 10 MVP detection rules / 10 条检测规则
│ ├── validators.go # Validators: ID Mod11-2, Luhn, USCC Mod31, Shannon entropy / 校验函数
│ └── classifier.go # Match → L1-L5 classification / 匹配结果分级
├── scanner/
│ ├── walker.go # Recursive walk + extension filter / 递归遍历 + 扩展名过滤
│ ├── extractor.go # Plain text/DOCX/XLSX/PPTX content extraction / 内容提取
│ └── pipeline.go # Concurrent scan pipeline / 并发扫描管道
├── server/
│ ├── server.go # HTTP routes + API handlers + false positive correction / HTTP 路由 + API
│ └── handlers.go # JSON/CSV output utilities / JSON/CSV 输出工具函数
└── ui/
├── embed.go # Resource embedding / 资源嵌入
├── index.html # Single-page application / 单页应用
├── app.js # Frontend logic / 前端逻辑
└── style.css # Styling + dark/light theme / 样式 + 暗色/亮色主题
| Rule / 规则 | Validation / 校验方式 | Level / 级别 |
|---|---|---|
| Chinese ID Card / 身份证号 | GB 11643-1999 Mod11-2 checksum | L3 |
| Phone Number / 手机号 | 1[3-9] prefix, 11-digit format | L3 |
| Bank Card / 银行卡号 | Luhn/Mod-10 algorithm (13-19 digits) | L3 |
| Email Address / 邮箱地址 | Standard email format | L2 |
| IPv4 Address / IPv4 地址 | Each octet 0-255 validation | L2 |
| USCC / 统一社会信用代码 | GB 32100-2015 Mod31 checksum | L3 |
| AWS Access Key | AKIA prefix, 20 characters | L5 |
| GitHub Token | ghp_ prefix, 36 characters | L5 |
| Private Key / 私钥文件 | PEM format private key block | L5 |
| High-entropy Secret / 高熵密钥 | Shannon entropy > 4.5 | L5 |
| Level / 级别 | Name / 名称 | Description / 含义 |
|---|---|---|
| L1 | Public / 公开 | No sensitive data found / 未发现敏感数据 |
| L2 | Internal / 内部 | Contains emails, IPs, etc. / 包含邮箱、IP 等一般性信息 |
| L3 | Confidential / 机密 | Contains IDs, phone numbers, bank cards / 包含身份证号、手机号、银行卡号等个人敏感信息 |
| L4 | Secret / 秘密 | 3+ L3 matches in the same file / 同一文件包含 3 项及以上 L3 级别数据 |
| L5 | Restricted / 管控 | Contains private keys, cloud API keys, high-entropy secrets / 包含私钥、云服务密钥、高熵密钥等核心凭证 |
Is my data safe? / 我的数据安全吗？
Yes. All scanning and analysis runs entirely on your local machine. No data is uploaded to any server. 完全安全。所有扫描和分析都在本地完成，不联网、不上传任何数据。
Which OS is supported? / 支持哪些操作系统？
Currently Windows only. The tool uses Windows-specific APIs for file system operations and browser launch. 目前仅支持 Windows，使用了 Windows 特有的文件系统操作和浏览器启动接口。
What is the maximum file size? / 文件大小限制？
50 MB per file. Larger files are automatically skipped. 单个文件最大 50 MB，超过此大小的文件会被自动跳过。