search-facebook

name	search-facebook
description	Use when scraping Facebook group rental listings using Chrome DevTools MCP tools — requires user to be logged into Facebook in their local Chrome browser

Facebook Group Rental Scraper (Chrome DevTools)

Scrapes rental listings from Facebook groups using Chrome DevTools MCP tools. Requires the user to be logged into Facebook in their local Chrome browser.

Prerequisites

• Chrome DevTools MCP server must be running and connected.
• User must be logged into Facebook in Chrome.
• Facebook groups to search are listed in config.yaml under facebook_groups.

Input

You will receive:

• city — city name in Hebrew (e.g., "הוד השרון")
• min_rooms — minimum number of rooms (e.g., 5)
• groups — array of Facebook group URLs from config

Scraping Process

Step 1: Check Chrome DevTools availability

mcp__chrome-devtools__list_pages

If this fails, return empty results with a note: "Chrome DevTools unavailable — Facebook skipped."

Step 2: Navigate to each Facebook group

For each group URL in the config:

mcp__chrome-devtools__navigate_page
URL: https://www.facebook.com/groups/XXXXXXXXX

Step 3: Verify login and access

mcp__chrome-devtools__take_screenshot

Check the screenshot:

• If a login page appears, report "Facebook login required — skipping" and return empty.
• If a "Join group" prompt appears, report "Group membership required for [group URL]" and skip to the next group.
• If the group feed is visible, proceed.

Step 4: Scroll to load posts

Scroll down multiple times to load recent posts:

() => { window.scrollBy(0, 1500); return 'scrolled'; }

Repeat 4-5 times. Facebook loads posts dynamically — take screenshots between scrolls to verify new content.

Step 5: Extract post content

Use evaluate_script to extract post text:

() => {
  const posts = document.querySelectorAll('[data-ad-comet-preview="message"], [role="article"]');
  return Array.from(posts).slice(0, 20).map(post => ({
    text: post.innerText?.substring(0, 1000) || '',
    links: Array.from(post.querySelectorAll('a')).map(a => a.href)
  }));
}

Step 6: Filter relevant posts

From the extracted posts, filter for ones that mention:

• The target city in Hebrew (e.g., הוד השרון)
• Room count keywords: 5 חדרים, 6 חדרים, etc. (>= min_rooms)
• Rental keywords: להשכרה, השכרה

Step 7: Extract details from relevant posts

For each relevant post, parse the text to extract:

• Address (look for street names)
• Price (look for ₪ or ש"ח followed by numbers)
• Room count (X חדרים)
• Size (X מ"ר or X sqm)
• Any amenities mentioned (חניה, מעלית, ממ"ד, מרפסת, מיזוג)
• Post permalink (extract from timestamp link or construct from post ID)

Use click on the post timestamp to get the permalink URL if needed.

Step 8: Normalize output

{
  "url": "https://www.facebook.com/groups/XXXXX/posts/YYYYY",
  "source": "Facebook",
  "city": "{city}",
  "address": "extracted from post text",
  "neighborhood": "",
  "rooms": 5,
  "floor": "",
  "size_sqm": null,
  "price": 8500,
  "mamad": null,
  "parking": null,
  "elevator": null,
  "ac": null,
  "balcony": null,
  "pets": null,
  "available_from": "",
  "contact": "extracted phone number if visible",
  "images": []
}

Rules

• Runs SEQUENTIALLY — shares Chrome browser with Madlan scraper.
• Posts are in Hebrew — parse Hebrew text for details.
• Facebook posts are unstructured — extract what you can, use null for missing fields.
• If post text is ambiguous about room count, include it if it might match (err on inclusion).
• Limit to 10 most recent matching posts per group.
• If a group is inaccessible (not a member, login required), skip it and note it in the output.

Output

Return the array of normalized listings from all groups. Report: groups searched, posts scanned, matching listings found. List any inaccessible groups separately.