🤖 AI Prompts

ShadowCopilot Prompts Library

Ready-to-use prompts for AI-powered web scraping

E-commerce Popular

Amazon Product Extraction

Extract the product name, current price, original price, discount percentage, rating (out of 5), number of reviews, availability status, and the first 5 bullet points from the product description.
Social Media

LinkedIn Profile Data

Extract the person's full name, headline, current company, job title, location, number of connections, and their last 3 work experiences with company names and durations.
Real Estate

Zillow Property Listing

Extract the property address, listing price, number of bedrooms, bathrooms, square footage, lot size, year built, property type, and Zestimate value if available.
Jobs

Indeed Job Listing

Extract the job title, company name, location, salary range if shown, job type (full-time/part-time), posted date, and the key requirements from the job description.
News

News Article Summary

Extract the article headline, author name, publication date, main image URL, article summary (2-3 sentences), and the 5 most important quotes from the article.
E-commerce

Product Reviews Analysis

Extract the first 10 reviews with: reviewer name, rating, review date, review title, review text, and whether it's a verified purchase. Also identify the top 3 positive and negative themes.
General

Contact Information

Find and extract all contact information from this page: email addresses, phone numbers, physical address, social media links (LinkedIn, Twitter, Facebook), and contact form URL if present.
SEO

Google Search Results

Extract the top 10 organic search results with: position, title, URL, meta description, and any featured snippets or rich results. Note if any results are ads.

💡 Prompt Writing Tips

✓ Do

  • • Be specific about what data you want
  • • Specify the output format (JSON, list, etc.)
  • • Include fallback instructions for missing data
  • • Use clear field names

✗ Don't

  • • Use vague terms like "get everything"
  • • Ask for data that requires login
  • • Request too many fields at once
  • • Forget to handle edge cases