doc-ocrSAFE

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

# 1) Agent Skills CLI (works with 50+ agent hosts)
npx skills add jiawood2006/doc-ocr

# 2) Claude Code / any host that reads ~/.claude/skills
git clone https://github.com/jiawood2006/doc-ocr.git ~/.claude/skills/doc-ocr

# 3) Hermes Agent
git clone https://github.com/jiawood2006/doc-ocr.git ~/.hermes/skills/utilities/doc-ocr

python3 scripts/dococr.py 合同.pdf                 # single file → text
python3 scripts/dococr.py 发票.jpg                 # image / scan
python3 scripts/dococr.py 书.pdf --md              # Markdown output
python3 scripts/dococr.py ./扫描件/ -o 全部.txt     # batch a folder

pip3 install pymupdf pyobjc-framework-Vision

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: doc-ocr
description: "文档文字识别。用户提供 PDF/扫描件/图片(合同、发票、书页、截图),需要提取文字、转成可编辑文本时使用。扫描件自动 OCR(macOS Vision,中英文)。Document OCR: extract editable text from PDFs, scans, and images (contracts, invoices, book pages, screenshots) via macOS Vision."
version: 1.0.0
author: 涛哥
license: MIT
metadata:
  hermes:
    tags: [ocr, pdf, document, text-extraction, scan]
    category: utilities
    homepage: https://gitee.com/tao6677/useful-tools
---

# Doc-OCR 文档文字识别

PDF / 扫描件 / 图片 → 可编辑文字。有文字层的 PDF 直接提取,扫描件自动 OCR(macOS Vision 自带,中英文)。

## 触发条件

用户提供 PDF/图片文件,要求:
- "提取文字""转文字""OCR"
- 处理扫描件、合同、发票、书页、截图

## 使用步骤

### 1. 单个文件

```bash
python3 ~/.hermes/skills/utilities/doc-ocr/scripts/dococr.py 合同.pdf
python3 ~/.hermes/skills/utilities/doc-ocr/scripts/dococr.py 发票.jpg
```

输出保存为 `<输入名>_ocr.txt`。

### 2. 批量目录

```bash
python3 ~/.hermes/skills/utilities/doc-ocr/scripts/dococr.py ./扫描件/ -o 全部.txt
```

### 3. Markdown 输出

```bash
python3 ~/.hermes/skills/utilities/doc-ocr/scripts/dococr.py 书.pdf --md
```

## 依赖(首次使用时安装)

```bash
pip3 install pymupdf pyobjc-framework-Vision
```

**注意**:OCR 依赖 macOS Vision(仅 macOS 可用)。Linux 需另装 tesseract 等引擎。

## 已知陷阱

- **扫描件判定**:PDF 文字层 <20 字自动走 OCR,正常 PDF 直接提取。
- **手写体**:Vision 对印刷体/清晰手写效果好,潦草手写不保证。
- **隐私卖点**:文件在本机处理,不上传第三方。

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

declared (1 observation(s))

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.