Redact — 技能 — openclaw中文资讯站

技能详情（站内镜像，无评论）

Privacy redaction toolkit for images, PDFs, Word documents, and PowerPoint presentations. Use when the user needs to redact, mask, or replace sensitive/priva...

媒体与内容

作者：noah @DarkNoah

许可证：MIT-0

MIT-0 ·免费使用、修改和重新分发。无需归因。

版本：v0.1.0

统计：⭐ 0 · 17 · 0 current installs · 0 all-time installs

⭐ 0

安装量（当前） 0

🛡 VirusTotal ：良性 · OpenClaw ：可疑

Package：darknoah/redact

安全扫描（ClawHub）

VirusTotal ：良性
OpenClaw ：可疑

OpenClaw 评估

The skill's code mostly matches a redaction toolkit, but there are incoherences and risky choices (an undocumented required 'uv' installer tool, unpinned runtime installs, and an explicit DISABLE_MODEL_SOURCE_CHECK setting) that warrant caution before use.

目的

The skill name/description (redaction for images/PDF/docx/pptx) aligns with the included scripts which perform OCR and redaction. However, the runtime initializer (scripts/init-runtime.sh) requires an external 'uv' binary (used to create venv and run pip) even though registry metadata lists no required binaries — that is an unexplained dependency and mismatch.

说明范围

SKILL.md and the scripts confine actions to local file processing (images, PDFs, docs, presentations). No hard-coded remote endpoints are present. However, every Python script sets DISABLE_MODEL_SOURCE_CHECK=True which suppresses model source verification, and the code uses Paddle/PaddleOCR which will typically download model weights at runtime — this expands the skill's scope to fetching and executing third-party models without verification.

安装机制

There is no formal install spec in registry metadata, but the provided init-runtime.sh attempts to install dependencies via 'uv pip install' (an uncommon installer wrapper). That means: (a) an undocumented external binary 'uv' is required; (b) pip installs unpinned packages (paddlepaddle, paddleocr, etc.) from PyPI which can be large and may pull additional binaries or wheels; (c) model weights for OCR may be downloaded at runtime. Using an un…

证书

The skill declares no required environment variables or credentials (which is appropriate), but the scripts forcibly set DISABLE_MODEL_SOURCE_CHECK='True' and FLAGS_use_mkldnn='0'. Setting DISABLE_MODEL_SOURCE_CHECK to True is notable: it weakens verification of model provenance and is not justified in the README, increasing risk of executing unverified model artifacts fetched from the network.

持久

The skill does not request always:true, does not claim to modify other skills, and only creates a local virtual environment under scripts/.venv. It does not request unusual persistence or system-wide configuration changes.

安装（复制给龙虾 AI）

将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。

请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「Redact」。简介：Privacy redaction toolkit for images, PDFs, Word documents, and PowerPoint pres…。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/darknoah/redact/SKILL.md
（来源：yingzhi8.cn 技能库）

SKILL.md

打开原始 SKILL.md（GitHub raw）

---
name: redact
description: |
  Privacy redaction toolkit for images, PDFs, Word documents, and PowerPoint presentations. 
  Use when the user needs to redact, mask, or replace sensitive/private information in files.
  
  Triggers:
  - Redacting or masking sensitive text in images, PDFs, documents, or presentations
  - Replacing names, phone numbers, IDs, or other PII in files
  - Processing privacy compliance for documents before sharing
  - Anonymizing content in visual files
  
  Supported formats: png/jpg images, PDF, docx/doc, pptx/ppt
---

# Redact Skill

Privacy redaction toolkit using PPStructureV3 OCR for text detection and replacement.

## Scripts

| Script | Format | Command |
|--------|--------|---------|
| `redact-image.py` | Images (png, jpg, etc.) | `redact-image.py <input> <rules.csv> <output>` |
| `redact-pdf.py` | PDF | `redact-pdf.py <input> <rules.csv> <output>` |
| `redact-document.py` | Word (docx, doc) | `redact-document.py <input> <rules.csv> <output>` |
| `redact-presentation.py` | PowerPoint (pptx, ppt) | `redact-presentation.py <input> <rules.csv> <output>` |

## CSV Rules Format

```csv
target_text,replacement_text
张三,李四
手机号,
身份证号,
```

| Rule | Effect |
|------|--------|
| `原文本,新文本` | Replace with new text |
| `原文本,` | Empty = mask with █ (documents) or solid color block (images/PDF) |

## Masking Behavior

| Format | Empty Replacement |
|--------|-------------------|
| Images, PDF | Solid color block overlay |
| Word, PowerPoint | `█` characters (same length as target) |

## Features

| Feature | Image | PDF | Document | Presentation |
|---------|:-----:|:---:|:--------:|:------------:|
| Text replacement | ✅ | ✅ | ✅ | ✅ |
| Solid color mask | ✅ | ✅ | - | - |
| █ character mask | - | - | ✅ | ✅ |
| OCR detection | ✅ | ✅ | ✅ (images) | ✅ (images) |
| Tables | - | ✅ | ✅ | ✅ |
| Headers/Footers | - | ✅ | ✅ | - |
| Embedded images | - | ✅ | ✅ | ✅ |

## Environment Setup

```bash
# Initialize virtual environment
./scripts/init-runtime.sh

# Activate
source scripts/.venv/bin/activate
```

## Dependencies

- Python 3.10+
- PaddleOCR / PPStructureV3
- python-docx, python-pptx, PyMuPDF, Pillow