Docs

Appearance

What's new ​

New products, parameters and plan changes, newest first. Every item links to the page that documents it in full.

November 2026 (6 November) ​

Airbnb Listing API: the stay price every time ​

  • Airbnb Listing API: a listing asked with dates now waits for its stay price, so price and price_text no longer come back empty on part of the requests, and the request takes about 6 s instead of 8-10 s.
  • Airbnb Listing API: a discounted stay now comes back with its price - price is the discounted price and price_text reads both, e.g. $423 for 3 nights, originally $486; these came back empty before.

Site APIs: a clear error for a site paused over oversized answers ​

  • Errors: a site API request to a site whose answers kept exceeding the size limit now ends at once with 502 TARGET_RESPONSE_TOO_LARGE and says the requests are paused for up to an hour, instead of TARGET_BLOCKED about a bot check. Nothing is charged, as before.

November 2026 (5 November) ​

Glassdoor jobs and companies ​

  • Glassdoor Listing API - search www.glassdoor.com jobs by keyword and city, state or country with salary, date posted, job type, company rating, remote and Easy Apply filters: 30 jobs a page with title, company and its rating, location, pay range with currency and period, age, Easy Apply and the sponsored flag; page on with cursor. MCP tool glassdoor_listing.
  • Glassdoor Job API - one job by its id or URL: the full description, pay range, job type, skills, education and experience, location with coordinates, apply link, review highlights and the employer's profile with CEO, website, size, revenue, founding year and ratings. MCP tool glassdoor_job.
  • Glassdoor Companies API and Glassdoor Company API - search companies by location, sector, industry and rating, or ask for one by id or URL: profile links, headquarters, size, industry, description, seven ratings, review, salary and job counts and office addresses. MCP tools glassdoor_companies and glassdoor_company.
  • The no-code Glassdoor Jobs Scraper sweeps up to 20 search pages into JSON, CSV or XML and can add each job's full details (include_detail).

Webhooks: one event per call, no-code runs, cache hits ​

  • Webhooks: a site API call sends one scrape.request.* with the credits of the whole call, instead of one per page it needed behind the scenes; a call that ended TARGET_BLOCKED sends scrape.request.failed with billingStatus: refunded and creditsRefunded.
  • No-code scraper runs send scraper.job.started, scraper.job.finished and scraper.job.failed (data.jobType: no_code_scraper), like the Website Crawler.
  • A cache hit sends scrape.request.finished (data.cacheHit: true), and a request that ran out of time sends scrape.request.failed with errorCode: SCRAPE_TIMEOUT.
  • The webhookUrl delivery of an async request that failed says reason: failed.

Page HTML from the real estate APIs ​

  • Redfin, Zillow and Apartments.com search and property APIs take include_html=true and also return the page the data was read from in html, as Yelp, Walmart and eBay do. The page is then listed with the request in Files too.
  • include_html=true is a paid add-on in every site API that takes it: +4 credits on top of the API's price (Billing and credits). The Playground shows it next to the option and in the cost preview.
  • Redfin Search API - search the homes of a city, ZIP code, neighborhood or county by name (location=Austin, TX) or by a redfin.com search URL, for sale, under contract or sold, with home type, price, beds, baths, size, lot, year built and HOA filters: up to 350 homes in one request with price, address, coordinates, facts, status, days on market, photo and open house, plus how many homes match. MCP tool redfin_search.
  • The no-code Redfin Scraper searches an area with the same filters and can add each home's full property page (include_detail); pasting property URLs or ids still works.

MCP results in Files, AI Assistant fixes ​

  • MCP server: a site API tool call keeps its JSON result in Files, as the same REST request does.
  • AI Assistant: a data answer now comes with its result file to download right under it; a new chat shows the suggested prompts by group; the assistant answers in the language of your question; "fill in a form" prepares the form card without running the request.

November 2026 (4 November) ​

Batch results as CSV from the API ​

  • Batch Scraping: GET /api/v1/scrape/batch/web/{jobId}/export?format=csv returns the status table of a batch - one row per URL with its status, billing, credits, target HTTP status and error - and status (for example status=failed) limits any export to the items in that status. The CSV button of the batch page downloads this file at once, even for a 50,000-URL batch, instead of collecting it page by page in the browser.

AI Assistant gets the data for you ​

  • AI Assistant: ask for data from a site ("get me the price of this product", "find coffee shops in Berlin") and the assistant runs the matching product API itself - the same tools the MCP server offers - then answers with the values, says what the run cost and offers the full result as a file to download in the chat, with a link to its record in Files.
  • AI Assistant: one question runs at most two charged calls, each a single request at the product's own price; a blocked or platform-side failure is not charged. A batch or a crawl is still never started from the chat - the assistant prepares it as a card and you start it on its page.

Files: a batch is one record ​

  • Files: a batch is one record instead of one per URL, with the whole batch to download as one file per format it was started with - JSON always, Markdown and Text when the batch asked for them. The item count and size of the record cover every item.
  • Files: every record has an Open run link to the page it came from - the batch page, the no-code scraper's own page or the request log - where its settings are and its results can be downloaded again.

No-code scrapers show their progress ​

  • No-code scrapers: a run reports pages_swept, pages_requested, failed_pages, total_items and credits_spent while it sweeps, so a script can show how far it is. The dashboard's Scrapers without code page shows the same live: the stage, pages swept of the page budget, records collected, credits spent, pages that brought nothing and how long the run has taken - instead of a disabled Run scrape button.

Tripadvisor search by city name, dates and currency ​

  • Tripadvisor Search API and the no-code Tripadvisor Scraper take a city by name: location=Lisbon, Portugal instead of looking up its geo_id.
  • Hotel partner prices for your stay with checkin, checkout and adults, and in currency: USD, EUR, GBP, CAD, AUD, INR, NZD or SGD.

Cost of a site API request ​

  • Every site API answer reports what the request charged in meta.cost.charged_credits, the way a Web Scraping API answer reports cost.

MCP server ​

  • MCP server - new tools for every site API: zillow_listing, zillow_property, apartments_listing, apartments_property, allegro_listing, allegro_offer, redfin_property, tripadvisor_search, tripadvisor_place, airbnb_search, airbnb_listing, yellowpages_search, yellowpages_business and amazon_best_sellers. Each takes the arguments of its REST endpoint, needs the scrape:web scope and costs the same.

Fixes ​

  • Tripadvisor Scraper API: a hotels or restaurants list returns all 30 places of the page; before, about one search in four returned 8 places, when Tripadvisor answered with the version of the page it serves to phones.
  • Yelp Scraper API: business pages and searches come back again. Requests rotate across residential exits in several countries instead of US exits, which Yelp's bot check started refusing; a page one country refuses is retried from another at no extra charge.
  • Google Maps Search API: a first results page ScrapeIt cannot read answers 502 TARGET_BLOCKED with block_reason: unrecognized_response and is not charged; before, it came back as a charged empty local_results.
  • Google SERP API and Google Maps API: gl=uk, Google's own code for the United Kingdom, works like gl=gb; before, it failed with 422 PROXY_PROVIDER_NOT_FOUND. A market ScrapeIt has no exit in now says to choose another gl instead of pointing at proxyType, which these APIs do not take.
  • MCP server: tools that scrape answer within 90 seconds like the REST site APIs; a call out of time answers SCRAPE_TIMEOUT and is not charged, instead of being cut off by the network after 100 seconds.
  • Airbnb, Booking.com and Tripadvisor APIs answer a date the calendar does not have (2027-02-30) or a check-in in the past with 400 VALIDATION_ERROR before anything is charged; before, the site was asked for a rolled-over or past stay and answered for other dates.
  • Google SERP API: search_metadata.request_id names the request the search was charged under; after a free second pass past a CAPTCHA it named that free pass, which the request log shows at 0 credits.
  • Website Crawler: /docs and /docs/ are one page, crawled and charged once; before, a site linking both forms had the page crawled twice.
  • Cache: a request with jsRendering: false and one with jsRendering omitted (auto) no longer share an entry, and captureWhenReady is part of the entry; before, an auto request could be served a fetched page it would have rendered, and a request waiting for a selector a page captured without that wait.
  • Credits come back when a site API request ends in 503 TARGET_DNS_UNAVAILABLE (our DNS resolver did not answer) after an earlier step was charged.

October 2026 (31 October) ​

Fixes ​

  • Credits come back when a site API request, a no-code run or a bulk run ends in an internal error after an earlier step was charged, and when a background no-code or bulk run is cut off by a restart of the service; before, those credits stayed charged.
  • Files: a Web Scraping API request answered from the cache is a record of its own, with the page and content it was served; before, it had no record because its page belonged to the request that filled the cache. A Web Scraping API request that asked for Markdown, text or extracted JSON keeps them in Files next to its HTML.
  • Files: the result files of a crawl or a no-code run, and a crawl's page HTML, are kept for your plan's window (raw_html_retention_hours) like every other file, instead of 6 months.
  • eBay Search API: delivery_estimate is filled from the shipping line when eBay writes the estimate there (Free delivery in 1-2 days gives Delivery in 1-2 days); before, it stayed null on search cards without a Get it by line.
  • Files: a Scraper API record offers the JSON answer; the page it was parsed from is listed only when the call asked for it with include_html.
  • Website Crawler and Batch Scraping explain what each is for and how they differ, in the dashboard and here: a batch scrapes the URLs you list, the crawler finds the pages of a site for you.
  • eBay Scraper API holds up under bursts of parallel requests: a refused attempt now retries through another exit right away instead of waiting on a browser, and a browser render is the last attempt of a request. A search with min_price, max_price or zip_code, and every page of the no-code eBay Scraper, stays on US exits, so the filters mean US dollars. With include_description, the request log shows one Item API step instead of two.
  • Google Maps Reviews API: a place id Google rejects answers 400 VALIDATION_ERROR instead of 502 TARGET_BLOCKED.
  • A paid site API answer is sent at once; its copy in Files is saved right after it, so a slow file storage no longer delays the response.
  • Batch export: an item whose content file is larger than 16 MiB is content: null in the export (download it by contentFileId), and every export is recorded in the activity log.
  • Batch export: an item whose content file cannot be read from storage is content: null and the export goes on; before, the download stopped there with a cut-off file.
  • Batch export in the dashboard: the JSON, Markdown and Text buttons of a batch page save the file straight to disk through a one-time link, so a large batch no longer has to fit in the browser tab's memory.
  • Website Crawler: a page whose Markdown or text was cut to 256 KB ends with a Truncated. note in markdown_results and text_results, naming the page's HTML file (htmlFileId) that holds all of it.
  • Markdown and Text cut out of a content JSON in Files work for content JSONs up to 16 MiB; a larger one answers PAYLOAD_TOO_LARGE - download it as JSON.
  • Website Crawler: a page result keeps only the links the crawl could follow (on a seed site and allowed by the path rules).
  • The request log sorted by status, credits or product with only an end date shows the 7 days before that date instead of an empty list.
  • Google Search API: the ai_overview text no longer falls apart into one character per line when a list item of the answer holds a cited-source link; cited-source rows are left out of the text.
  • Amazon Product API: rating_histogram is returned when a star row shows 0% (Amazon draws it without a link); that star reads 0 instead of the whole histogram being null.
  • Batch Scraping in the dashboard: a workspace member with the developer role can stop a batch; before, Stop batch answered "permission denied". The button now asks for confirmation and explains that unstarted URLs are skipped and not charged.
  • The dashboard keeps you signed in when the service does not answer for a moment, for example during an update: the request asks you to try again, and a page opened at that moment offers Try again instead of the sign-in form. Before, you were signed out and lost what you had typed.
  • API keys belong to the member who created them: removing a member from the workspace revokes their keys, and a role change revokes the keys with scopes the new role does not hold. A developer can revoke or rotate only their own keys; an owner or an admin can manage every key. The dashboard warns about it before you remove a member or change their role.
  • A max_price below min_price is now refused with 400 VALIDATION_ERROR before any credit is charged, instead of a billed empty page, on the Amazon, Walmart, eBay, Zillow, Allegro and Apartments.com Scraper APIs. The Walmart Search API also refuses a brand that contains |.
  • Files: the download token of GET /api/v1/files/{id}/download is accepted only in the x-download-token header, so it never lands in proxy logs or browser history. A request with ?token= and no header returns 400 VALIDATION_ERROR and the token stays unused - send it in the header instead.
  • The dashboard asks you to sign in again at least once every 90 days: staying active no longer extends a session past 90 days from the sign-in that started it.
  • After 10 wrong passwords in a row, signing in to that account is paused for 15 minutes (each next pause in a row doubles, up to 24 hours); the answer is 429 RATE_LIMIT_EXCEEDED with Retry-After. A successful sign-in resets the count, and resetting the password lifts the pause at once.

October 2026 (30 October) ​

eBay Scraper API ​

  • New eBay Scraper API: eBay Search API (GET /api/v1/scrape/ebay/search) returns one page of 60, 120 or 240 www.ebay.com listings with price and currency, condition, buying format, bids and auction end time, shipping, location, sold count, watchers, badges and the sponsored flag, plus the applied filters, item-specific filters, related searches and pagination. Filters: price, condition, buying format, free shipping, US-only, category, results per page and a ZIP code for shipping costs.
  • eBay Item API (GET /api/v1/scrape/ebay/item) returns one item by number or URL: the price in the listing's own and the shown currency, condition and its description, item specifics, every variation with its own price and stock, the seller with feedback, shipping, returns, payment methods, auction state, photos, category and reviews; include_description adds the seller's full description at no extra charge.
  • The no-code eBay Scraper sweeps up to 20 search pages into JSON, CSV or XML, optionally with every item page under detail. Both APIs are also MCP tools (ebay_search, ebay_item). 15 credits per request (introductory).

Google Maps Reviews API ​

  • Google Maps Reviews API - the reviews of one place in the order you choose (most relevant, newest, highest or lowest rating first), optionally only those that mention a keyword: rating, text and Google's translation, date, likes, photos, visit details such as price per person and food or service ratings, the owner response and the reviewer. 15 credits per request. Google shows a signed-out visitor up to five reviews per order and keyword; another order or keyword returns another set. Also in the Playground and as the google_maps_reviews MCP tool.

Files: one record per request ​

  • The dashboard's Files page shows one record per request instead of one row per file, named by the product that made it (Airbnb Search API, Web Scraping API, Batch, Website Crawler…). Each record has a button for every format of its result: JSON, Markdown, Text, HTML, screenshot, CSV, XML - and a product filter.
  • Scraper API calls now keep the JSON they answered with in Files, next to the page it was parsed from, for the plan's retention window.

AI Assistant sets up every product ​

  • The AI Assistant now knows the whole live catalog - every Scraper API (Google, Amazon, Walmart, Zillow, Redfin, Airbnb, Yellow Pages, Tripadvisor, Booking.com, Yelp, Allegro, Apartments.com), the Web Scraping API, batch scraping, the Website Crawler and the no-code scrapers - with their parameters and prices.
  • Ask for a search, a scrape, a crawl or a batch and the card it prepares shows complete cURL, Node.js and Python code for the request, and opens that product's Playground (or the Batch or Crawler page) with the form filled in. Parameters are checked before the card appears, so the form opens with valid values.
  • New suggested prompts cover the site APIs, code examples, batch and crawl.

Richer data from the existing site APIs ​

The same requests now return more of what each page carries - no extra credits, no slower answers.

  • Amazon Scraper API - unit price, coupon, bought in the past month, delivery, other offers, subtitle and the total result count on search; parent ASIN, rating histogram, badges, delivery, returns, videos, A+ content, specifications, dimensions, weight, every variation and frequently bought together on a product.
  • Google SERP API - the AI Overview with its references, the local pack, top stories, ratings and sitelinks on organic results, profiles and the exact review count in the knowledge panel.
  • Google Maps API - the time zone of every place; on a place: address components, plus code, reservation link, photo count, rating distribution, review topics and highlights, secondary hours.
  • Walmart Scraper API - unit price, seller, fulfillment, badges, categories and related searches on search cards; return policy, warranty, rating counts, review highlights and the reviews shown on a product.
  • Yelp Scraper API - the review snippet and hours on search cards; summary, claimed status, year established, history, health inspection, popular dishes and review highlights on a business.
  • Zillow Scraper API - open houses, unit, price changes, showcase flag, rental phone, highlights, available units, base rent and required monthly fees on listings; time zone, full and half baths, parking, home insights and the virtual tour on a property.
  • Apartments.com Scraper API - total results, the market snapshot, required monthly fees and the lease term.
  • Booking.com Scraper API - deals, price badges, location score, unit and beds, partner programs and sustainability on search results; highlights, child policies, deposit, payment methods, guest questions, featured reviews, prepayment terms and the tax breakdown on a property.
  • Allegro Scraper API - price with delivery, seller rating, recent buyers, offer count, condition and the Smart! flag on listings; the delivery group and the seller's negative ratings on an offer.

Batch and crawl results in one file ​

  • Batch Scraping - GET /api/v1/scrape/batch/web/{jobId}/export?format=json|markdown|text merges the content of every item into one file. The batch page in the dashboard saves JSON, Markdown, Text and the CSV status table, plus each item's HTML and JSON, without a detour through Files.
  • Website Crawler - a crawl that asked for markdown or text also gets markdown_results and text_results files with every page. The crawler page saves every result file and each page's HTML directly, and its status updates by itself until the files are ready.

New filters ​

  • Google SERP time_range (hour, day, week, month, year).
  • Amazon search department, min_price, max_price.
  • Walmart search min_price, max_price, brand, in_stock_only, sold_by_walmart. Sponsored results are walmart.com ads it places regardless of sold_by_walmart; check sponsored and seller_name.

Website Crawler ​

  • The crawler form in the dashboard opens with an example you can run at once - ten product pages of books.toscrape.com over the plain fetch - and every field shows what it does, an example value and its default.
  • JavaScript rendering starts off in the form, as it already was for API requests without jsRendering: pages load over the plain HTTP fetch, several times cheaper. Turn it on for sites that build their content with JavaScript.

Fixes ​

  • Website Crawler follows every link of a page again: a crawl with a small limit only looked at the first limit links, and when those were excluded by excludePaths or includePaths (category pages, the home page) it stopped after the first page.
  • Yelp Scraper API business pages answer again: Yelp now draws the page in the browser, and the request waits for it and its reviews instead of reading an empty page as a block. On a page Yelp serves without its data cache, the business id, categories, website, claimed status, open state, year established, history, health score and related searches are read from the page itself.
  • Allegro Scraper API listings return the item condition alone (Używany), not the whole parameter line.
  • While a platform data store briefly does not answer (for example during a restart), requests get 503 SERVICE_UNAVAILABLE with a Retry-After header instead of 500 INTERNAL_ERROR or a request that hangs until your client times out - see errors. The SDKs retry it automatically. A Scraper API request that ends in this error costs no credits.
  • The request log sorted by status, credits or product answers in a fraction of a second on large accounts: without date_from it sorts the last 7 days (sorting).
  • The Amazon Product API example ASIN is a regular product page.
  • The dashboard Playground runs the Amazon Best Sellers, Redfin, Airbnb, Yellow Pages and Tripadvisor APIs, and every product form opens filled with a working example, ready to run - the Web Scraping API one with a cheap plain fetch of https://books.toscrape.com/. Every field shows its type, an example and its default value.

October 2026 (29 October) ​

New site APIs ​

No-code scrapers ​

  • Amazon (search, products, Best Sellers), Google (search, Maps, Maps reviews), Redfin, Airbnb, Tripadvisor and Yellow Pages run without code from the dashboard or over POST /api/v1/scrapers/{scraper}/jobs. The Amazon search, Google Maps, Airbnb, Tripadvisor and Yellow Pages scrapers take include_detail to nest each result's own page.

The Amazon Best Sellers, Redfin, Airbnb, Tripadvisor and Yellow Pages prices are introductory and may change.

October 2026 (28 October) ​

Errors ​

  • A non-existent target hostname answers 422 TARGET_HOST_NOT_FOUND instead of 403 SSRF_BLOCKED; a resolver that briefly cannot answer gives a retryable 503 TARGET_DNS_UNAVAILABLE. See Errors.

October 2026 (22-24 October) ​

Plans ​

  • New plans. Free (3 000 credits a month, two requests at a time, webhooks included), Starter (€35), Pro (€79), Business (€169) and Ultra (€449); annual billing is 20% cheaper. Launch, Scale and Enterprise are no longer sold - running subscriptions keep their terms. See Billing & credits.

Web Scraping API ​

  • jsRendering: "auto" is the default: the request starts on the fast fetch and is rendered only when the page needs a browser. See auto rendering.
  • AI extraction with aiExtractRules: describe the fields, get JSON back, on paid plans. See Structured data extraction.
  • Richer extractRules: attributes, HTML output, lists and nested rules. New jsScenario steps: press, select and evaluate. See data extraction and page interactions.
  • Proxy and session control: proxyType: "auto", sticky sessions with sessionId, and waitUntil. See proxy options.
  • Async requests: POST /api/v1/scrape/web/async returns a job id at once and posts the signed result to your webhookUrl, with receiver examples in eight languages. See Async requests.

New site APIs ​

The Google Maps, Walmart, Booking.com and Yelp prices are introductory and may change.

API keys ​

  • The Website Crawler, the no-code bulk jobs and the jobs API accept API keys with the scrapers:run, scrapers:read and scrapers:stop scopes. These scopes were removed from existing keys, where they granted nothing before: create a new key with them for crawler or bulk-job automation. See Authentication & API keys.

Integrations ​

Reliability ​

  • A public status page with live success rate and response times per product, plus GET /api/v1/public/status.

ScrapeIt Web Scraping API documentation. Managed web scraping services: www.scrapeit.io.

© ScrapeIt. All rights reserved.