Install
openclaw skills install @zhaoxinghua09-cell/doc-desens-scanner当用户说『这篇文档有敏感信息要发出去』『先脱敏再发』『别把内部路径/密钥/人名露出去』『对外发布前扫一遍隐私』,或要把文档/代码/日志对外前做去敏时使用。扫描 个人标识(PII)·密钥·内部路径·内部项目代号 四类敏感项并打码/留痕,输出去敏版+清单(有证:脱敏动作可审计)。可运行脚本(desens_scan 扫描器)。理论根基:LGD 三律之有证(脱敏留痕,可追责)。触发词:去敏、脱敏、desensitize、脱敏扫描、隐私打码、密钥泄露、内部路径、对外发布前检查、PII。
openclaw skills install @zhaoxinghua09-cell/doc-desens-scanner凡自治之物 · LGD 治理学 · 创造术家族 | 评测协议:见平台质量门禁规范
Use it when the user plans to share a document, piece of code or log externally; when content might contain internal paths or leaked secrets and needs checking; and when a pre-release privacy-compliance self-check is required to make sure no personal identifiers (PII) or other sensitive data goes out.
Call the core script python desens_scan.py with the target file path, e.g. python desens_scan.py -i document.txt. To customize the desensitization rules, add --target-key [type] (e.g. pii, key, path). The system reads the input content directly and scans / processes it.
Step 1: the user provides the raw document or code stream as input. Step 2: the engine loads desens_scan.py and parses the content in depth, identifying the four sensitive classes: PII, secrets, internal paths and internal codenames. Step 3: automatic masking (redact / replace) runs and a traceable processing record is generated. Step 4: the redacted content plus a full risk list is produced. Step 5: the audit log is attached when delivering the result, satisfying the theoretical root of "evidenced — accountable".
用户计划将文档、代码或日志对外共享时;发现内容可能存在内部路径、密钥泄露风险需要排查时;在发布前执行隐私合规自检,确保不包含个人标识(PII)及其他敏感信息时使用。
调用核心脚本命令 python desens_scan.py,并传入目标文件路径参数 [path],例如 python desens_scan.py -i document.txt。若需自定义脱敏规则,可添加参数 --target-key [密钥类型](如 pii, key, path);系统将直接读取输入内容进行扫描与处理。
第一步:用户提供待处理的原始文档或代码流作为输入。第二步:技能引擎加载 desens_scan.py 脚本对内容进行深度解析,识别 PII、密钥、内部路径等四类敏感项。第三步:执行自动脱敏操作(打码/替换)并生成可追溯的处理记录。第四步:输出包含完整风险清单的脱敏版本内容。第五步:交付最终结果时附带审计日志,满足“有证可追责”的理论根基要求。