Brightdata

What it does

This skill provides access to Brightdata's Mosaic, a global web data network, enabling AI agents to scrape and fetch content from websites that might be blocked or difficult to access directly. It allows for both single URL scraping/fetching and crawling multiple pages from a site. The skill requires an initial voice notification before any action is taken.

When to use it

• When you need to retrieve content from a website that's blocking your direct access.
• To scrape or fetch a specific URL.
• To crawl multiple pages from a website, whether for mapping the site or extracting data.
• When encountering bot detection or CAPTCHA issues while accessing a webpage.
• For comprehensive content extraction beyond basic scraping capabilities.

Key capabilities

• Access to Brightdata's Mosaic web data network.
• Single URL scraping/fetching using a four-tier progressive scraping workflow (WebFetch → Curl → Browser Automation → Bright Data MCP).
• Website crawling, with options for light crawls (under 50 pages) and full crawls via the Bright Data Crawl API.
• Workflow routing based on user requests (e.g., single URL vs. site crawl).
• Mandatory voice notification before any action is performed.

Example prompts

• "Scrape this URL: [URL]"
• "Crawl all pages under /docs on example.com"
• "I can't access this site; use Bright Data to fetch it."
• “Get me the content from [URL]”

Tips & gotchas

• Mandatory Notification: You must send a voice notification using the provided curl command immediately upon invoking the skill. This is not optional.
• Workflow Selection: The agent will automatically route your request to either the "FourTierScrape" workflow for single URLs or the "Crawl" workflow for crawling multiple pages.
• Customization: User-specific configurations can be found and applied from ~/.claude/PAI/USER/SKILLCUSTOMIZATIONS/BrightData/.