Site Adapters

Documentation Index

Fetch the complete documentation index at: https://web2md.org/docs/llms.txt

Use this file to discover all available pages before exploring further.

​

Overview

While Web2MD works on any webpage, certain sites benefit from specialized extraction logic. Web2MD includes 16 platform-specific extractors that use each platform’s API or DOM structure to produce better Markdown output than generic extraction.

​

Supported sites

​

API-based (no tab needed)

These extractors fetch data directly via the platform’s API, so they work even without an active browser tab.

​

Reddit

Extracts the full post content including:

• Post title, author, subreddit, and score
• Self-text or link content
• Up to 200 comments with 4 levels of nesting, including author and score
• Image and media links

Uses Reddit’s JSON API (/.json) for reliable extraction.

​

Hacker News

Extracts full discussion threads via Algolia API:

• Post title, URL, points, and comment count
• All comments with author and nesting

​

arXiv

Extracts research papers:

• Paper title, authors, and abstract
• Full HTML content for research papers

​

YouTube

Extracts video metadata:

• Video title and channel name
• Duration and view count
• English subtitles/captions (when available)

​

Twitter / X

Extracts tweet content:

• Tweet text and media
• Author info and engagement metrics

​

DOM-based (requires browser tab)

These extractors parse the page DOM inside the active tab for structured content.

​

OpenAI Docs

Extracts documentation pages from OpenAI’s documentation site:

• Code blocks and API references
• Structured navigation and content hierarchy

​

Claude Chat

Extracts conversation content from the Claude.ai chat interface.

​

Notion

Extracts content from public and logged-in Notion pages (notion.so, notion.site):

• Dual extraction: block-level DOM parsing (data-block-id) with container fallback
• Headings, text, code blocks, lists, toggles, callouts, quotes
• Images, tables, bookmarks, and embeds

​

Feishu / Lark

Extracts content from Feishu Docx and Wiki pages (feishu.cn, larksuite.com):

• Dual extraction: PageMain block model with DOM fallback
• Headings, text, code blocks, lists, tables, images
• Callouts and checkboxes

​

Medium

Extracts article content from Medium’s DOM:

• Author, publication date, and reading time
• Full article content

​

Substack

Extracts newsletter posts:

• Author, date, and full article content

​

Wikipedia

Extracts article content:

• Sections, tables, and references
• Removes navigation and sidebar clutter

​

DEV.to

Extracts developer blog posts:

• Tags, reactions, and comments

​

Product Hunt

Extracts product pages:

• Product descriptions and maker info
• Discussion threads

​

Stack Overflow

Extracts Q&A content:

• Question title, body, tags, and vote count
• Accepted answer (highlighted)
• Top answers with vote counts

​

GitHub

Extracts Issues and Pull Requests including:

• Title, status (open/closed/merged), and labels
• Original description/body
• Up to 30 comments with author info

Uses GitHub’s REST API for structured data.

​

How adapters work

When you convert a page, Web2MD automatically:

1. Detects the site from the URL
2. Selects the best adapter if one exists
3. Extracts content using the site-specific method
4. Falls back to generic extraction if the adapter fails

You don’t need to configure anything — adapters are applied automatically.

​

Comparison