PaddleOCR agent skills

Install

pip install "paddleocr>=3.7.0"
npx skills add PaddlePaddle/PaddleOCR -g --skill paddleocr-text-recognition -y
npx skills add PaddlePaddle/PaddleOCR -g --skill paddleocr-doc-parsing -y

This is third-party code. Review the repository files before installing.

What it does

The paddleocr-text-recognition skill extracts line-level text with bounding boxes and confidence from screenshots, photos, scans and PDFs. The paddleocr-doc-parsing skill handles complex layouts: tables, formulas as LaTeX, charts, seals, multi-column text and reading order, returning Markdown or JSON. Both skills call the paddleocr api command, which uses the PaddleOCR cloud API, and tell the agent which skill to pick and how to handle errors.

Who it is for. For people who need an agent to pull text and tables out of scans, invoices, reports and scientific PDFs.

Example request

Parse this report PDF and return the main text and all tables as Markdown

Limitations

Requires Python 3.9+, paddleocr 3.7.0 or newer and a PADDLEOCR_ACCESS_TOKEN from Baidu AI Studio. Files are processed by the cloud API and quotas apply. Availability of AI Studio and its sign-up from Russia is unverified.

How to disable. Delete the paddleocr-text-recognition and paddleocr-doc-parsing folders from the agent's skills directory and remove PADDLEOCR_ACCESS_TOKEN from settings.

Security check

• Sends documents to the PaddleOCR cloud API
• Uses an AI Studio access token

README in short

The skills documentation explains how to choose between the two skills, the requirements (Python, the paddleocr package, an AI Studio token) and three install options: the skills CLI, clawhub for OpenClaw and manual copying. It shows how to set PADDLEOCR_ACCESS_TOKEN in .claude/settings.local.json and in OpenClaw, with example prompts. The skills live in the skills folder of the main PaddleOCR repository, Apache-2.0 licensed.

SKILL.md

---
name: paddleocr-doc-parsing
description: >-
  Use this skill to extract structured Markdown/JSON from PDFs and document images—tables with
  cell-level precision, formulas as LaTeX, figures, seals, charts, headers/footers, multi-column
  layout and correct reading order.
  Trigger terms: 文档解析, 版面分析, 版面还原, 表格提取, 公式识别, 多栏排版, 扫描件结构化,
  发票, 财报, 复杂 PDF, PDF转Markdown, 图表, 阅读顺序; reading order, formula, LaTeX,
  layout parsing, structure extraction, PP-StructureV3, PaddleOCR-VL.
license: Apache-2.0
metadata:
  openclaw:
    requires:
      env:
        - PADDLEOCR_ACCESS_TOKEN
      bins:
        - paddleocr
    primaryEnv: PADDLEOCR_ACCESS_TOKEN
    install:
      - kind: uv
        package: paddleocr
        bins: [paddleocr]
---

# PaddleOCR Document Parsing

## When to Use This Skill

**Use this skill for**:

- Documents with tables (invoices, financial reports, spreadsheets)
- Documents with mathematical formulas (academic papers, scientific documents)
- Documents with charts and diagrams
- Multi-column layouts (newspapers, magazines, brochures)
- Complex document structures requiring layout analysis

## Usage

### Basic Document Parsing

From URL:

```bash
paddleocr api \
  --model_type doc_parsing \
  --file_url "https://example.com/report.pdf"
```

Related