Craigslist Scraper: Multi-City Listings Monitor, Only New

Search Craigslist in many cities at once and get one row per listing: title, price and currency, posting time, city, neighborhood, coordinates, category, bedrooms and square feet for housing, pay for jobs, image links and the listing link. Give it up to 20 keywords and up to 100 cities, or paste up to 100 Craigslist search URLs with your filters already set. Turn on only new and schedule it: each run returns just the listings it has not delivered before. Turn on listing details to add what only the posting page shows: year, make and model, odometer, condition and title status for vehicles, bathrooms, laundry, parking and pets for rentals, employment type for jobs.

Built for three jobs:

• Resellers and deal hunters watching the same items (bikes, phones, tools, furniture) across many cities every hour or day.
• Rental and housing analysts collecting asking rents and locations in the cities they cover.
• Dealers and lead researchers following used cars, equipment or job posts in their markets.

The Actor does not collect phone numbers or email addresses. If a poster typed one into a title, it is replaced with [contact removed].

Quick start: one keyword, two cities

{
  "searches": ["iphone"],
  "cities": ["newyork", "losangeles"],
  "maxResultsPerSearch": 50
}

You get up to 100 rows (50 per city), newest first, plus a RUN_SUMMARY record with a status for every search.

What it costs

$1.20 per 1,000 listings ($0.0012 per delivered listing) plus $0.001 per run start. Platform usage is included. Searches that fail, are refused or find nothing cost nothing beyond the start.

Listing details are optional (includeDetails): +$0.50 per 1,000 listings whose posting page loaded ($0.0005 each), so 1,000 listings with details cost $1.70. A page that is gone or refused costs nothing extra; the listing itself is still delivered.

Run	Listings	Price
Quick start above	up to 100	up to $0.121
10 cities x 3 keywords, 200 each	up to 6,000	up to $7.201
Hourly only-new check, 20 new listings	20	$0.025
Hourly only-new check, nothing new	0	$0.001 (about $0.72 a month)
Used cars in 3 cities, 100 each, with details	up to 300	up to $0.511

Set Maximum cost per run in the run options to cap a bill. When the cap is reached, the run stops cleanly, keeps every row already delivered and marks the searches it did not finish as CHARGE_LIMIT.

Inputs

Field	What it does
searches	Keywords, up to 20 (queries and keywords also work). Every keyword runs in every city. Leave it empty to list the newest listings of the category.
cities	Craigslist sites, up to 100 (locations and areas also work): newyork, sfbay, chicago, a sub-area such as newyork/mnh, a host such as seattle.craigslist.org or a site name such as new york city.
category	Craigslist category code, sss (all for sale) by default: for example cta cars and trucks, apa apartments, bia bicycles, ela electronics, jjj jobs.
searchUrls	Craigslist search URLs copied from your browser, up to 100. Their filters are kept as they are.
minPrice, maxPrice, postedToday	Filters added to every keyword search (a search URL's own filter wins).
maxResultsPerSearch	1 to 360, 200 by default. Craigslist returns at most the 360 newest matches per search.
onlyNew	Return only listings that appeared since the previous run of this saved task (or of your account, without a task).
includeDetails	Also open each delivered listing's page for its attributes (off by default). One extra request per listing, about 0.3 s each: 1,000 listings take about 5 minutes, so give large runs a longer timeout. With onlyNew, only new listings are opened.
proxyConfiguration	Apify Proxy (datacenter) by default.

A city Craigslist does not know, a malformed URL or an unsupported category becomes an INVALID_INPUT row in the summary; the rest of the run continues.

Run it on a schedule

Save your input as a task, switch on onlyNew and add a schedule (hourly or daily). The first run returns the newest listings (up to maxResultsPerSearch) and remembers the rest of the current results as seen; every later run returns only what appeared since then. If a run stops at your maximum cost, the listings it could not deliver come in the next run. Each saved task keeps its own memory, so two tasks never hide listings from each other. If two runs of the same task overlap, the second one stops with MONITOR_BUSY instead of returning duplicates. Runs with different searches can overlap safely, even in the same city.

Output

One row per listing, newest first within each search:

Field	Example
postingId	7951015478
url	https://www.craigslist.org/view/d/emeryville-trek-road-bike/...
title	Trek road bike 56cm
price, currency	380, USD (empty when the poster gave no price)
postedAt	2026-11-01T20:07:38Z
city, area, neighborhood	sfbay, eby, Emeryville
latitude, longitude	37.8404, -122.2894 (approximate, as shown on the Craigslist map)
category, categoryName	bik, bicycles - by owner
bedrooms, sqft	1, 700 on housing listings (0 bedrooms is a studio; empty when the poster did not say)
compensation	$20 per hour, the pay as a job or gig listing states it
imageUrls	links to the listing's photos (600x450)
search	the keyword or search URL that found the listing
firstSeenAt	when the listing was first delivered (only-new mode)
scrapedAt, fingerprint, schemaVersion	observation time, a hash of the listing's fields, the row format version

With includeDetails, each row also gets (the Listings with details view shows them):

Field	Example
year, makeModel	2017, Ford Escape SE (vehicles)
odometer	59300 (miles, as a number)
condition, titleStatus	excellent, clean
bathrooms	1.5 (housing)
attributes	every attribute the page lists, for example {"fuel": "gas", "transmission": "automatic", "VIN": "...", "laundry": "w/d in unit"}
detailsStatus, detailsError	OK (charged as details); GONE (removed since the search), BLOCKED, FAILED or NOT_REQUESTED (run timeout near or after repeated refusals) are not. With details on, a run with a maximum cost delivers only the listings it can pay for with their details

Measured on 2026-11-02 across 660 pages: year, make and model, odometer and title status on every car listing; bathrooms on 98% of apartments.

A listing found by two of your searches is delivered and charged once.

RUN_SUMMARY in the run's key-value store lists every search with its status and row count:

Status	Meaning	Charged
OK	Listings delivered	per listing
NO_RESULTS	Craigslist answered and has no matching listings	no
INVALID_INPUT	Unknown city, bad URL or a filter Craigslist rejected	no
SOURCE_FAILED	Craigslist did not answer correctly after retries	no
BLOCKED	Craigslist refused the request; the city's other searches are skipped	no
CHARGE_LIMIT	Your maximum cost was reached before this search finished	delivered rows only

The run fails only when every search failed or was refused. A run with some failed searches succeeds and says which ones to retry.

Use with AI agents through Apify MCP

Add this Actor to an MCP client such as Claude or Cursor:

https://mcp.apify.com?tools=automa-flow/craigslist-scraper

Example prompt:

Run automa-flow/craigslist-scraper with searches ["trek", "specialized"],
cities ["sfbay", "portland", "seattle"], category "bia", maxPrice 600 and
maxResultsPerSearch 100. Show the 20 cheapest listings with price, city,
postedAt and url, and list any search whose status is not OK. Keep
maxTotalChargeUsd at 1.00.

An agent can predict the bill before the run: $0.001 per run plus at most $0.0012 per listing ($0.0017 with includeDetails), and at most cities x searches x maxResultsPerSearch listings.

Sources and limitations

• Data comes from Craigslist's public search, the same results its website shows, read with one request per search (plus one per listing with includeDetails) at a polite pace. This Actor is not affiliated with craigslist.
• Craigslist's terms of use restrict automated access. You are responsible for how you use the data; do not use it to contact posters in bulk.
• Each search returns at most the 360 newest matches. For more, split the search by sub-area, price range or category.
• When a city has only a few matches, Craigslist adds listings from nearby cities. The Actor keeps only the city (and sub-area) you asked for and counts the others as nearbySkipped in RUN_SUMMARY; they are not charged. Add those cities to cities if you want them.
• Listing descriptions, phone numbers, email addresses and reply links are not collected. Listing details read only the page's attribute list; a contact typed into an attribute is replaced like in titles.
• Craigslist can change its search or refuse requests. The Actor then reports SOURCE_FAILED or BLOCKED and charges no listing (only the $0.001 start); it never fills in listings.
• onlyNew remembers the newest 1,000 listings per search and forgets a search that has not run for 30 days. A listing the seller renews moves back to the top of Craigslist; it counts as new only if this search never returned it before.

Release history

See the CHANGELOG.md.