web-scrape

name	web-scrape
description	Scrape web pages to clean markdown with optional AI summaries
version	0.1.0
tools	["Bash"]

Web Scrape

Fetches web pages, converts HTML to clean markdown, with optional AI summarization.

When to Use

• User wants to save a web article as markdown
• Batch scraping a list of URLs
• Building a research corpus from web sources

Commands

# Scrape a single URL
bun run web-scrape/scripts/scrape.ts --url "https://example.com/article"

# Scrape with AI summary
bun run web-scrape/scripts/scrape.ts --url "https://example.com/article" --summarize

# Batch scrape from file (one URL per line)
bun run web-scrape/scripts/scrape.ts --file urls.txt

# Batch scrape with summaries
bun run web-scrape/scripts/scrape.ts --file urls.txt --summarize

Requirements

• GEMINI_API_KEY environment variable (only for --summarize)
• Output goes to output/scraped/ by default

Output

Markdown file with YAML frontmatter (title, author, source, word count) and clean article content. The --summarize flag prepends bullet-point summary + topic tags.

Features

• Extracts <article> or <main> content when available
• Strips nav, footer, scripts, ads
• Preserves headings, links, code blocks, lists
• Extracts metadata from Open Graph and meta tags