phy-pinterest-scraper_skill

Overview

This skill scrapes Pinterest search results to collect pins with direct pin URLs, highest-resolution image links, and descriptions without requiring login. It uses a headless Playwright browser with infinite-scroll handling to gather results reliably. Output is saved as a JSON file keyed by a safe version of the search query for easy downstream consumption.

How this skill works

The scraper navigates to Pinterest search pages, waits for pin elements, and repeatedly scrolls to trigger infinite loading. After each scroll it runs in-page JavaScript to extract pin IDs, resolve srcset values to the highest-resolution image URL, and capture alt text or descriptions. Results are deduplicated by pin ID and the process stops when the requested maximum is reached or scrolling stalls.

When to use it

• Collect visual inspiration or reference images for design and creative work
• Assemble image datasets for machine learning or aesthetic analysis
• Monitor trends or popular imagery in a niche for competitive research
• Automate mood board generation or content planning workflows
• Capture product imagery from search results for cataloging or research

Best practices

• Set a sensible max_pins limit and add pauses between queries to avoid rate limits
• Run headed (headless=False) while debugging DOM or selector mismatches
• Increase max_scroll_attempts or scroll_delay if fewer results appear than expected
• Respect rate limiting guidelines and avoid scraping large volumes from one IP
• Handle CAPTCHAs and network errors by retrying with backoff or rotating IPs

Example use cases

• Gather 50 mood-board images for a branding project using a single search term
• Build a dataset of product images for computer vision training with consistent resolution
• Scan competitor-related queries to see which visual styles dominate search results
• Automate collection of interior design images for trend analysis over time
• Quickly extract high-resolution images and descriptions for content planning