article-extractor

name	article-extractor
description	Download HTML from websites and extract article content, removing HTML bloat. Specifically designed to extract content from <article> tags with aria-live="polite" attribute. Outputs clean, readable content for documentation sites like Red Hat docs.
author	Gabriel McGoldrick ([e-mail])
allowed-tools	Read, Bash, Write

Article Extractor Skill

This skill downloads HTML from websites and extracts the article content, removing unnecessary HTML bloat. It's particularly useful for documentation websites that have large amounts of navigation, styling, and other non-content HTML.

Capabilities

• Download HTML: Fetch HTML content from any publicly accessible URL
• Extract Article Content: Locate and extract content from <article> tags
• Clean Output: Remove unnecessary HTML bloat and format for readability
• Multiple Output Formats: Support for HTML, Markdown, and plain text output
• Flexible Matching: Find article tags by various attributes (aria-live, class, id)

Usage

The skill uses a Python script that downloads and parses HTML content.

Basic Usage

Extract article from a URL:

uv run --script ${CLAUDE_SKILL_DIR}/scripts/article_extractor.py --url "https://example.com/page"

Extract with specific output format:

uv run --script ${CLAUDE_SKILL_DIR}/scripts/article_extractor.py --url "https://example.com/page" --format markdown

Save to file:

uv run --script ${CLAUDE_SKILL_DIR}/scripts/article_extractor.py --url "https://example.com/page" --output article.md

Extract with custom article selector:

uv run --script ${CLAUDE_SKILL_DIR}/scripts/article_extractor.py --url "https://example.com/page" --selector "article.main-content"

Command Line Options

• --url URL: The URL to fetch HTML from (required)
• --format {html,markdown,text}: Output format (default: markdown)
• --output FILE: Save output to file instead of stdout
• --selector SELECTOR: CSS selector for article content (default: article[aria-live="polite"])
• --pretty: Pretty-print HTML output with indentation
• --strip-links: Remove all hyperlinks from output

Output Formats

HTML (default): Extracts the article HTML content with all tags preserved but removes surrounding bloat.

Markdown: Converts the article content to Markdown format for easy reading and documentation.

Plain Text: Strips all HTML tags and returns plain text content.

Examples

Red Hat Documentation

# Extract from Red Hat OpenShift Lightspeed documentation
uv run --script ${CLAUDE_SKILL_DIR}/scripts/article_extractor.py \
  --url "https://docs.redhat.com/en/documentation/red_hat_openshift_lightspeed/1.0/html/install/ols-installing-lightspeed" \
  --format markdown \
  --output openshift-lightspeed-install.md

Generic Documentation Site

# Extract from any site with article tags
uv run --script ${CLAUDE_SKILL_DIR}/scripts/article_extractor.py \
  --url "https://example.com/docs/guide" \
  --selector "article.documentation" \
  --format text

Dependencies

This skill requires the following Python packages:

• requests: For downloading HTML content
• beautifulsoup4: For parsing and extracting HTML
• html2text: For converting HTML to Markdown (optional, for markdown format)

Install dependencies:

python3 -m pip install requests beautifulsoup4 html2text

Use Cases

1. Documentation Archival: Extract and save documentation pages for offline reading
2. Content Migration: Extract article content for migration to different platforms
3. Clean Reading: Get rid of navigation, ads, and other distractions
4. Content Analysis: Extract main content for text analysis or processing
5. Markdown Conversion: Convert HTML documentation to Markdown format

Performance

The skill downloads and processes HTML efficiently:

• Single page extraction: ~1-3 seconds depending on page size and network
• No rate limiting by default (respect target site's robots.txt)
• Memory efficient streaming for large pages

Limitations

• Only works with publicly accessible URLs (no authentication support)
• Requires the target page to use <article> tags or similar semantic HTML
• JavaScript-rendered content may not be extracted (uses static HTML only)
• Some complex HTML structures may not convert perfectly to Markdown