Vibeleaderboard
← Back to Tools
Index / tool

Huashu MD/HTML

https://github.com/alchaincyf/huashu-md-html
Visit github.com
Category
AI Agents
Rank
No. 971Tools index
Pricing
Open Source
Type
TOOL
Builder
alchaincyf
GitHub
901 stars
Date

About

Claude Code skill wrapping markitdown + Pandoc + html-to-markdown + trafilatura into a two-way pipeline between Markdown and beautiful HTML, with anti-AI-slop themes.

What it does

Huashu MD/HTML treats Markdown as the editable master copy for a document workflow. It imports common office files, media, archives, and web pages into Markdown, then produces themed HTML or formatted Word documents. Published HTML can also return to Markdown for editing and archiving.

Why it's ranked here

This is a thoughtfully scoped publishing toolkit, not merely a format converter. It distinguishes readable articles from structured web pages, offers separate output styles for articles, reports, reading, and long documents, and includes a Word path built for editorial handoff. Its value depends on accepting several external tools and occasional conversion cleanup.

What's good

The URL routing guidance is unusually practical: preserve fields and hierarchy for reference pages, but remove navigation and advertising from articles. HTML themes use restrained typography, responsive layouts, dark modes, and local CSS. Word output adds covers, contents, images, tables, chapter styling, headers, footers, and page numbers.

Tradeoffs

There is no graphical interface. Installation spans Python packages plus a separate Pandoc binary, and the declared dependency list omits the packages required for Word output. Scanned PDFs need an LLM or Azure service for useful OCR. Slides lose animation and layout, complex tables can lose meaning, and web extraction does not bypass login walls or anti-scraping controls.

How to use it well

It suits writers and technical teams who keep Markdown under version control, publish the same source in several forms, or hand Word drafts to editors. Choose extraction based on whether a page is meant for reading or reference. Expect to edit imported Markdown. Use other tools when exact layout restoration, spreadsheet analysis, slides, PDF production, or custom visual art is the real requirement.

Technical notes+

The repository exposes four Python CLI entry points: scripts/any_to_md.py, scripts/md_to_html.py, scripts/html_to_md.py, and scripts/md_to_docx.py. scripts/any_to_md.py wraps MarkItDown and optionally configures OpenAI image descriptions or Azure Document Intelligence. scripts/html_to_md.py fetches URLs with urllib.request, applies Trafilatura extraction, then selects html-to-markdown or markdownify. scripts/md_to_html.py invokes Pandoc through subprocess.run, loads CSS and optional Pandoc templates from templates/, supports copied or base64-inlined images, and enables a table of contents for the interactive theme. Its VALID_THEMES also contains wechat, although the main documentation presents four HTML themes. requirements.txt declares MarkItDown, html-to-markdown, Trafilatura, and markdownify, but not the python-docx and Pillow dependencies imported by scripts/md_to_docx.py. The documented repository tree contains no test directory.

Observed

License
MIT License, permitting personal and commercial use, modification, distribution, sublicensing, and sale.
Primary language
Python, with four command-line conversion scripts.
Install surface
Installs as an agent skill through npx skills add; Python dependencies and Pandoc must also be installed separately.
Interfaces
Agent skill plus command-line interfaces; no buttons, panel, or GUI.
Platform support
Pandoc installation instructions cover macOS, Debian or Ubuntu, and Windows.
Repository structure
Includes scripts, HTML templates, reference documentation, examples, demos, and a requirements file; no test directory appears in the supplied tree.

Read from README.md, requirements.txt, scripts/any_to_md.py, scripts/html_to_md.py, scripts/md_to_docx.py, scripts/md_to_html.py, LICENSE, SKILL.md, examples/readme.html, references/anti-ai-slop.md, references/design-tokens.md, references/md-to-html-themes.md, references/html-to-md-cookbook.md, references/markitdown-cookbook.md, references/md-to-docx-cookbook.md.

What it can do

  • Convert Markdown to HTML

    Markdown text → HTML document

  • Convert HTML to Markdown

    HTML document → Markdown text

  • Extract content from web pages

    Web page URL or HTML → Clean Markdown text

  • Apply anti-AI-slop themes to HTML

    HTML document → Styled HTML with theme

  • Process documents through Pandoc conversion

    Various document formats → Converted document format

  • Clean and format extracted web content

    Raw web page content → Structured Markdown text

Tags

claude-codeskillmarkdownhtmlpandoc

Tech Stack

Python

Comments (0)

No comments yet

Editorially curated, with community endorsements as a secondary signal. Corrections welcome.