Apartment Hunter
An apartment-search agent that scrapes Facebook groups and Yad2 daily against saved criteria and emails a filtered, deduplicated digest.
The Need
Apartment hunting meant checking eight Facebook groups and running two saved Yad2 searches every day, manually screening new listings and tracking which ones I had already reviewed.
Process
A Python scraper pulls listings from selected Facebook groups and Yad2 against saved criteria (city, size, price, safe room and other filters), with a layered approach to anti-bot protection: direct API calls first, then cookie-authenticated requests, then a real Chrome browser with a persistent profile to get past Yad2’s bot detection.
Listings are deduplicated against a local SQLite database so the same apartment never appears twice, then compiled into a daily HTML email digest with links to each one. The scraping and filtering logic is also packaged as an MCP server, so the same tools can be called directly from Claude Desktop or AntiGravity.
Value
Replaced a daily manual check across nine sources with a single email digest, removing a recurring daily task entirely. Qualitative only, no time-tracked metric.
Tech
Live, in personal use. Code is open source on GitHub; a few friends have asked to run it themselves with their own credentials.