libtransform

libtransform——资源转换实用工具。pdfToHtml 可利用 LLM 的视觉能力，将 PDF 文档转换为 HTML，将每一页拆分为图像，并通过多模态模型进行处理。适用于文档转换、PDF 处理，以及从文档中提取知识。

SKILL.md

--- frontmatter

name: libtransform
description: >
  libtransform - Resource transformation utilities. pdfToHtml converts PDF
  documents to HTML using LLM vision capabilities, splitting pages to images and
  processing with multimodal models. Use for document conversion, PDF
  processing, and knowledge extraction from documents.

libtransform Skill

When to Use

•Converting PDF documents to HTML
•Extracting content from scanned documents
•Processing documents with LLM vision models
•Building document transformation pipelines

Key Concepts

pdfToHtml: Splits PDF into page images, sends to vision-capable LLM, and assembles HTML output with semantic structure.

Usage Patterns

Pattern 1: Convert PDF to HTML

javascript

import { pdfToHtml } from "@copilot-ld/libtransform";

const pdfBuffer = await fs.readFile("document.pdf");
const html = await pdfToHtml(pdfBuffer, {
  model: "gpt-4-vision-preview",
  maxPages: 50,
});

Integration

Used by libingest pipeline for document processing. Requires LLM with vision capabilities.