跳到正文
TakoAPI
返回技能列表

docling

由 er3mit4 提供

Web & Frontend Development
868下载量由 clawskills.sh 提供
0浏览量
0 个赞登录后即可为这个技能点赞
—暂无评分
你的评分

简介

Extract and parse content from web pages, PDFs, documents (docx, pptx), and images using the docling CLI with GPU.

描述

该技能的功能

Docling is a CLI tool that extracts and parses content from web pages, PDFs, and office documents into clean, structured text. It supports OCR on scanned documents and images using GPU acceleration via CUDA.

Unlike basic HTTP fetchers, docling handles PDFs, scanned images, and complex document formats that return raw binary or malformed HTML.

何时使用

  • Extracting readable text from a research paper PDF
  • Converting a web article to markdown for offline processing
  • Running OCR on a scanned invoice or image
  • Parsing a PPTX slide deck into plain text
  • Pulling structured content from a DOCX report

示例工作流

输入

User asks: extract the content from this PDF: /tmp/report.pdf

智能体
  1. 1

    Run docling /tmp/report.pdf --ocr --device auto --output /tmp/docling_out

  2. 2

    Wait for docling to process the document with OCR

  3. 3

    Read the output markdown file from /tmp/docling_out

  4. 4

    Return the parsed text content to the user

  5. 5

    Clean up /tmp/docling_out

输出

Clean, structured markdown text extracted from the PDF

前置要求

  • docling CLI installed (e.g. via pipx install docling)
  • NVIDIA GPU with CUDA drivers (optional, for GPU-accelerated OCR)

安装

clawhub install er3mit4/docling

运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。

链接