ClawHub
Other
#paddleocr#ocr#document-parsing#pdf-to-markdown

PaddleOCR OCR & Document Parsing Setup

Install and configure two PaddleOCR Agent Skills for text recognition and structured document parsing in Codex, Claude Code, GitHub Copilot, Cursor, OpenCode, OpenClaw, and other compatible agents. Use for OCR and image-to-text from screenshots, photos, scans, and PDFs; Chinese/CJK text and bounding boxes; PDF-to-Markdown/JSON; tables, formulas, layout, and reading order; or endpoint, token, installation, and troubleshooting help.

Aidenwu0209@aidenwu0209

Install

openclaw skills install @aidenwu0209/paddleocr-skills-setup

Set up PaddleOCR Skills

Install and verify the two skills from the linked repository. Treat this setup skill as an installer guide; run OCR with the installed target skills.

Use this setup skill when the request mentions PaddleOCR installation or Agent Skills, or asks for 图片转文字 / 截图识字 / 扫描件 OCR / PDF 转 Markdown / 表格提取 / 公式识别 / 版面分析. It installs:

  • paddleocr-text-recognition for screenshots, photos, scans, PDFs, CJK text, line-level OCR, and optional bounding boxes.
  • paddleocr-doc-parsing for structured Markdown/JSON, tables, formulas, figures, multi-column layout, and correct reading order.

Install

  1. Confirm the user wants a global, user-level installation.

  2. Check Node.js/npx, Python 3.9+, and uv. Explain any missing prerequisite before using its official installer. Do not use sudo or change unrelated system settings without permission.

  3. Run:

    bash
    npx skills add Aidenwu0209/PaddleOCR-Skills --skill '*' -g -y

  4. Run npx skills list -g --json and confirm both paddleocr-text-recognition and paddleocr-doc-parsing, including their actual install paths.

Configure safely

  • Use https://www.paddleocr.com for the official API and token flow.
  • Ask the user to provide the OCR or document-parsing endpoint only when that target skill needs it.
  • Never invent, print, or store PADDLEOCR_ACCESS_TOKEN in chat, command history, source files, or logs.
  • Report incomplete configuration instead of claiming the OCR service works.

Finish

Report prerequisite versions, commands executed, installed skill names and paths, and any configuration still required from the user.