Scavio

E-commerce SEO

v0.25.3

Last updated Oct 14, 2026

n8n community node for Scavio — real-time structured JSON from Google, ChatGPT, Gemini, Amazon, Walmart, eBay, Target, Home Depot, Costco, Mercado Libre, YouTube, Reddit, TikTok, Instagram, X, LinkedIn, Threads, Kuaishou, Zillow, Redfin, Realtor.com, Righ

1,493 Weekly Downloads

2,210 Monthly Downloads

View on NPM GitHub Repository

Included Nodes

Scavio

Description

n8n-nodes-scavio

This is an n8n community node. It lets you use Installation – Operations – What changed in 0.13.0 – Per-resource notes – Templates – Credentials – Compatibility – Resources

Installation

Follow the installation guide in the n8n community nodes documentation.

In n8n: Settings -> Community Nodes -> Install -> enter n8n-nodes-scavio.

Operations

Search, retail, and marketplaces

| Resource | Credits | Operations |
| — | — | — |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>ChatGPT | 5 | Prompt |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>Amazon | 1 | Search Products, Get Product, Get Offers |
| source=n8n-node&utmmedium=package&utmcampaign=readme”>Get Stores |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>Target | 1 | Search Products, Get Category Products, Get Product, Get Reviews |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>Costco | 1; Get Warehouse Prices, Get Availability and Get Gas Prices scale with the number of warehouses | Search Products, Get Category Products, Get Categories, Get Products, Get Warehouse Prices, Get Availability, Get Reviews, Get Warehouses, Get Gas Prices, Get Coupon Book, Get Deals, Find Clearance Items, Get Search Suggestions |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>YouTube | 1; Search and Search Shorts 2, Get Streams 3, Get Transcript 5 | Search, Search Shorts, Search Channels, Get Suggestions, Get Video, Get Metadata, Get Comments, Get Comment Replies, Get Transcript, Get Related, Get Channel, Get Channel Videos, Get Channel Shorts, Get Channel Community, Resolve Channel, Get Streams, Get Community Post, Get Community Post Comments |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>TikTok | 1 | Get Profile, Get User Posts, Get Video, Get Video Comments, Get Comment Replies, Search Videos, Search Users, Get Hashtag, Get Hashtag Videos, Get User Followers, Get User Followings, Get Trending Searches, Get Trending Videos, Get Search Suggestions, Get Music, Get Music Videos, Search Music, Get Music Chart, Get Similar Creators, Get User Playlists, Get Playlist Videos, Get User Live Status, Get User Region |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>Instagram | 1-2: Get Profile, Get User Reels, Get User Tagged and Get Music Posts 1, the rest 2 | Get Profile, Get User Posts, Get User Reels, Get User Tagged, Get User Stories, Get User Highlights, Get Highlight Stories, Get Similar Accounts, Get Post, Get Post Comments, Get Comment Replies, Get Post Likers, Search Users, Search Hashtags, Search Reels, Search Music, Get Music Posts, Search Locations, Get User Followers, Get User Followings |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>LinkedIn | 1, 10, or 30 by operation | Get Person, Get Person About, Get Person Posts, Get Company, Get Company Posts, Search Jobs, Get Job, Get Post, Get Post Comments |
| source=n8n-node&utmmedium=package&utm_campaign=readme”>Kuaishou (China) | 1, 2, 10, or 40 by operation | Get Profile, Get User Posts, Get User Live Status, Resolve Share Link, Get Video, Get Video Comments, Get Comment Replies, Get Videos Batch, Search All, Search Videos, Search Users, Search Live Streams, Get Tag Feed, Get Trending |

Real estate and travel

App stores, software, and business reviews

What changed in 0.25.3

ChatGPT Prompt operation description now says 2 credits, matching the notice above the form.

What changed in 0.25.2

ChatGPT Prompt now costs 2 credits per answered prompt. Gemini stays at 5.

What changed in 0.25.1

ChatGPT Prompt now costs 1 credit per answered prompt (was 5). Gemini stays at 5.

What changed in 0.25.0

New resources: ChatGPT and Gemini, one Prompt operation each, 5 credits. Send a prompt and get the answer the assistant gives, for tracking which brands and pages AI assistants recommend. Both return one shape: answer.markdown, sources[] (ChatGPT adds title, publisher and date to each), search_queries[] (the web searches ChatGPT ran) and shopping[] (the products the answer recommends).
ChatGPT takes a prompt up to 1024 characters plus optional Model (default auto), Country (default us) and Include HTML. Gemini takes a prompt up to 4000 characters.
Answers typically take 10-40 seconds; both operations carry a 120-second request timeout. A failed prompt is not charged. A 503 means capacity is busy: wait for Retry-After and retry.

What changed in 0.24.0

Google: Custom Search (CSE Format). Returns the Google Custom Search JSON API response (items[] with title, link, snippet, displayLink; searchInformation.totalResults; queries.nextPage), so workflows built on Custom Search keep their downstream nodes. Takes the Custom Search parameter names: q, num (1-10), start (1-based), gl, hl, lr, cr, safe, dateRestrict, siteSearch + siteSearchFilter (enforced on the results), exactTerms, excludeTerms, orTerms, fileType. 1 credit. For the full SERP (knowledge graph, AI Overview, People Also Ask) keep using Search.

What changed in 0.23.1

Field descriptions say what each operation returns. No price, parameter or behavior changed.

What changed in 0.23.0

Instagram now costs 1-2 credits per call instead of 8-10, and gained 8 operations. No field or operation was renamed; existing Instagram nodes keep running and simply cost less.

Prices. Get Profile 1 credit. Get User Reels and Get User Tagged 1 (2 the first time a username is looked up). Get User Posts, Get User Stories, Get Post, Get Post Comments, Get Comment Replies, Get User Followers and Get User Followings 2 per call or page. Search Users and Search Hashtags 2, or 8 when you send a Cursor. Any Instagram call can occasionally cost more, up to 10, when a request needs extra work; the response’s credits_used is the exact charge.
Stories and highlights. Get User Highlights lists a user’s story highlights (ID, title, cover, item count); Get Highlight Stories returns every story in one highlight with image and video URLs. 2 credits each.
Discovery. Search Reels (keyword, cursor-paginated), Search Music (song or artist, returns the Music ID Get Music Posts takes), Get Music Posts (the Reels that use a track plus its total reel count, 1 credit per page), Search Locations (place ID, name, address, coordinates) and Get Similar Accounts (most creator accounts; some brand accounts return none). 2 credits each unless stated.
Engagement. Get Post Likers returns up to about 1,000 accounts that liked a post, plus its total like count, by Shortcode or URL. 2 credits.

What changed in 0.22.0

Extract gained 7 operations: map a website, crawl it, and extract a list of URLs in one job. Extract Page behaves as before; only its Mode option wording changed. Nothing else changed.

Map Website returns a site’s page URLs from its sitemaps and its start page, deduplicated, with optional Search text, include / exclude path globs, subdomains and a link cap (Max Links, sent as limit; API default 1,000; Free 500, PAYG 1,000, paid plans 5,000). Each link says whether it came from the sitemap or the page, with lastmod when the sitemap has one; total_found and truncated say whether there were more. 1 credit.
Start Crawl extracts same-site pages breadth-first from one URL within Max Pages (API default 100), Max Depth (0-10, default 3) and path globs, and returns a job ID. URLs under the start path are crawled first. Start Batch extracts a list of URLs (one per line or comma-separated) as one job; invalid URLs are listed in the response and never billed.
Billing is per successful page at the Extract price for the mode: Normal 1 credit, Advanced (renders JavaScript) 1, Ultra (the hardest pages) 2. Failed and skipped pages are free. Credits are taken page by page and a job stops cleanly with stoppedreason: insufficientcredits if the balance runs out.
Plan limits (max pages per crawl / active jobs / URLs per batch): Free 10/1/10, PAYG 100/1/50, Project 500/2/100, Bootstrap 2,000/3/500, Startup 5,000/5/1,000, Growth 10,000/10/1,000. Starting a job past your active-job limit returns 429 with Retry-After.
Get Crawl Status / Get Batch Status return progress (discovered, queued, succeeded, failed, skipped), credits used and the page results paged with next_cursor (Results Per Page 1-100, API default 25), optionally filtered by status or without content. Cancel Crawl / Cancel Batch stop a job; results so far stay available, and a finished job answers 409. Status and cancel are free. Results are kept for 3 days after a job finishes (2 MB per page, 250 MB per job); a job runs for at most 6 hours.
Wait for Completion (off by default) on Start Crawl and Start Batch polls the job inside the node, every Poll Interval seconds (default 5) for up to Max Wait seconds (default 300, max 600), then returns the status and first page of results. If the wait runs out, the latest status comes back with wait.timedout: true instead of an error; carry on with Get Status. The start response’s fields, including webhooksecret, are kept under created. Large jobs are better run with the toggle off plus a Wait node, or a webhook.
robots.txt is always obeyed for crawl and map. Batch reads each URL like Extract Page.
Webhook URL (optional) receives a signed POST when a job completes, fails or is cancelled, retried after 30 seconds and 5 minutes. Verify the Scavio-Signature: t=,v1= header: HMAC-SHA256 of . with the webhook_secret returned once by the start operation. Idempotency Key makes a repeated start within 24 hours return the original job.

What changed in 0.21.0

19 new operations on TikTok, YouTube, Reddit and X, 1 credit each. YouTube Get Transcript now costs 5 credits (was 8). Nothing else changed.

TikTok, 12 operations. Get Trending Searches (about 90 terms, US-weighted) and Get Trending Videos (about 6 videos per call whatever Count says; run again for a fresh sample), neither with a region filter. Get Search Suggestions (long-tail keywords can come back empty and still bill). Sounds: Get Music, Get Music Videos (about 14 per page; Count is ignored upstream), Search Music and Get Music Chart (Top 50 or Viral 50, one global chart). Creators: Get Similar Creators (about 30 per page, page with Page Token), Get User Playlists and Get Playlist Videos (feed it the mixId). Get User Live Status (live room status 2 = live, 4 = not live; an unknown username is a billed 404) and Get User Region (the registered region, not nationality, plus secuserid; an unknown username is an unbilled 404).
These TikTok operations return TikTok’s own payload. Large numeric IDs can lose precision when read as JSON numbers, so map the string fields (idstr, secuid, secuserid) in later nodes.
YouTube, 2 operations. Get Community Post (text, author, likes, images and a comments_cursor) and Get Community Post Comments.
Reddit, 2 operations. Search Comments (each result carries the post and subreddit it was left on; sort by relevance, hot, top or new) and Search Subreddits (subscriber counts, NSFW flags).
X, 3 operations. Search Communities, Get Community (description, member count, topic, rules) and Get Community Tweets (relevance or recency).

What changed in 0.20.0

New resource: Mercado Libre, 5 operations, 2 credits each (Get Reviews 1). Plus one description fix on Realtor.com.

Search Listings covers Mercado Libre in Mexico, Argentina and Colombia, and Mercado Livre in Brazil (Country; Chile is not supported). Pick Search By: keywords, a numeric seller ID (combines with keywords, filters, sort and paging), or a seller’s public nickname (their first page of listings, unfiltered, plus the seller ID to go further). Filters for condition, min / max price in the local currency, minimum discount, free shipping, Full fulfillment, official stores and interest-free installments; sort by relevance or price only (Mercado Libre has no newest or best-selling sort). About 48 listings per page and up to 42 pages, about 2,000 results per query; slice by price range, condition or seller to reach more. Rows carry price, list price and discount, installments, shipping flags, a sponsored flag, rating, units sold as a range (“+500”) and catalog IDs; seller name appears only where Mercado Libre shows it on the card.
Get Product takes a catalog product ID, an item (listing) ID or a URL (Look Up By) and returns price, promotions, the buy-box seller with reputation, stock, shipping, variations, specs, the star distribution and categories. Only the buy-box seller is returned; the full list of sellers on a product is not available.
Get Reviews returns 15 reviews per page, with an optional Stars filter (1 credit). Reviews translated by Mercado Libre keep the original text. Get Questions returns buyer questions with the seller’s answers (answers are dated, questions are not); page with Offset from next_offset. Get Best Sellers returns the top 20 in a category for Mexico, Argentina or Colombia; not available for Brazil, and the category ID prefix must match the country.
Brazil and Argentina may apply a “new” condition on their own for some queries; the response lists it in autoappliedfilters. A query with no results is an empty, billed 200. An unknown product, item or seller nickname answers 404 naming what was not found, billed like a normal call; so does a listing with no public questions.
Realtor.com Keywords: the description now says what the filter does: a listing must match all the keywords, so each one narrows the results (one Portland search: workshop 421, garage 2,764, both together 338). Behaviour is unchanged.

What changed in 0.19.0

Two new resources: Realtor.com, 6 operations, and Rightmove (UK), 7 operations, 1 credit each. Nothing else changed.

Realtor.com Search Listings covers for-sale (default; includes pending and contingent, as on the site), pending, for-rent and sold listings. Pick Search By: a location (city, county, ZIP, state or neighborhood), coordinates plus a 0.1-50 mile radius, or one agent’s listings. Filters for price, beds, baths, square feet, lot size, year built, garage, HOA, property types, listed / sold / price-cut within N days, Keywords matched in the listing text (comma-separated, up to 10), new construction, foreclosures and open houses, with 10 sorts. Up to 200 listings per call (Page Size, default 50). A search reaches its first 10,000 results, so Page x Page Size may not exceed 10,000; slice big areas by price band, property type or radius.
Sold prices are null in US non-disclosure states (Texas, Utah, Wyoming, Alaska, Idaho, Kansas, Louisiana, Mississippi, Montana, New Mexico, North Dakota and parts of Missouri). Sold dates still come back.
Realtor.com Get Property takes a property ID or a realtor.com URL (Look Up By) and returns price and tax history, three home value estimates with history and forecast, schools, flood / wildfire / noise risk, the listing agent and a mortgage estimate. Estimates, schools and risk can be null for land or new construction. Get Location Suggestions returns the slug_id that Get Market Stats takes (median list, sold and rent price, days on market, year-over-year change, by property type). Search Agents finds agents by city, ZIP code or name (one of the three, up to 50 per page); Get Agent Profile returns one agent with every review and four sub-ratings.
Rightmove Search Listings covers UK property for sale or to rent. Pick Search By: a locationid from Look Up Locations, a place name (resolved to its first match and echoed back), or an agent branch. Filters for price (GBP; monthly rent on To Rent), beds, baths, property types, must-haves, exclusions, tenure, added within 1-14 days, under offer, radius, furnishing and let type, with 4 sorts. 24-25 per page, up to 42 pages (about 1,000 listings) per search. There is no keyword filter on Rightmove; match on summary and keyfeatures in the results instead.
Rightmove Get Sold Prices returns 25 addresses per page in England and Wales, each with its full sale history since 1995, plus an area summary; filter by property type, tenure, sold within 1-30 years and radius, up to 40 pages. Get Sold Property returns one address’s every sale and its archived last listing. Get Property returns a live listing in full (tenure and lease, council tax band, service charge, EPC, floorplans, nearest stations); live listings carry no price history. Search Agents (20 branches per page, up to 50 pages) and Get Agent Branch (by branch ID or profile URL) round it out.
Property Types, Must Have, Exclude and Furnishing are multi-selects; Realtor.com Keywords takes comma-separated values.
An unknown property, agent, sold address or branch ID answers 404 naming what was not found, and is billed like a normal call. A removed Rightmove listing is a billed 404 too.

What changed in 0.18.0

New resource: Trustpilot, 6 operations, 2 credits each. Nothing else changed.

Search Businesses finds businesses by keyword in one country, up to 100 per page: name, domain, TrustScore, stars, review count, categories, website and address. Email and phone appear only when the business published them.
Get Business Profile takes a domain or a Trustpilot URL (pick with Look Up By) and returns the whole profile: rating distribution, review languages, claimed and verification status, contact details, reply rate and days to reply on negative reviews, consumer alerts, similar businesses and the 20 newest reviews. The AI summary and topics come back in Language (default en) when Trustpilot has them for that language, null or empty otherwise.
Get Reviews returns 20 reviews per page, filtered by stars, language (default all), date range, topics, text search, verified only and with replies, sorted by recency or relevance. totalfiltered counts every match, so a date range doubles as review velocity. Trustpilot serves up to 10 pages (200 reviews) per filter combination; slice by stars, language, date range, topics or text search to reach more. maxpage in the response says how far the current combination goes.
Get Categories returns the full category tree (22 top-level categories and their subcategories), or the categories matching a name. Get Category Businesses ranks businesses in one category and country, 20 per page, with sort, Min TrustScore (3.0+, 4.0+ or 4.5+) and claimed-only filters. Trustpilot applies Min TrustScore to the star rating, which rounds the TrustScore, so a 3.8 passes 4.0+.
Get Review returns one review by its 24-character review_id, with the business reply and the business it is about.
Stars and Topics take comma-separated values. Topic IDs come from Get Business Profile’s topics.
An unknown domain, review ID or category ID answers 404 naming what was not found, and is billed like a normal call. There is no TrustScore history and no reviewer-profile operation.

What changed in 0.17.1

Every request now identifies the node with User-Agent: n8n-nodes-scavio/ and X-Client-Source: n8n-nodes-scavio. No operation, parameter or output changed.

What changed in 0.17.0

New resource: Costco, 13 operations. Nothing else changed.

Search Products, Get Category Products and Get Deals share one set of filters: sort, brands, price range, minimum rating, on-sale, in-stock, page and page size. Set Warehouse ID to add that warehouse’s in-store price, stock status and price code to every result.
Get Warehouse Prices compares the online price with the in-warehouse price for up to 20 items at up to 10 warehouses: discount, final price, promotion start and end dates, per-order and per-member limits and price code. 1 credit for up to 9 warehouses, 2 at 10.
Get Availability returns in-warehouse stock status for one item at up to 10 warehouses, plus pickup and same-day delivery. Status only, never quantities. 1 credit per 5 warehouses.
Get Gas Prices returns regular, premium and diesel prices near a ZIP code or coordinates, or at specific warehouses. By location 1 credit (2 on Costco.ca); by warehouse IDs 1 credit per 5 warehouses (2 per warehouse on Costco.ca).
Get Coupon Book returns the current member coupon book as data: item number, amount off, final price, scope, limits and valid dates. United States only, no parameters.
Find Clearance Items scans a query or category at one warehouse for markdown price codes: .97 clearance and .00/.88 manager markdown by default, .49/.79/.89 special buy on request. The codes are the member-community decode; Costco does not publish them, and only items Costco lists online are covered.
Get Products, Get Reviews, Get Categories, Get Warehouses and Get Search Suggestions round out the catalog. Get Warehouses costs 2 credits on Costco.ca.
Item IDs, Warehouse IDs and Brands take comma-separated values.
Country is United States (default) or Canada everywhere except Get Coupon Book. Search Products, Get Products and Get Warehouses also accept the United Kingdom, Australia, Mexico, Japan, South Korea and Taiwan, with keyword, sort and paging only.

What changed in 0.16.0

Walmart store targeting is back, and Walmart gained a Get Stores operation. This reverses the 0.13.0 note below that Delivery ZIP and Store ID were gone.

Get Stores takes a ZIP / Postal Code (a 5-digit US ZIP, or a Canadian postal code such as M5V 2T6 with Domain set to Walmart.ca) and returns the nearby stores, nearest first, with store_id, distance, address, coordinates and hours. 1 credit.
Search Products and Get Product accept Delivery ZIP and Store ID under Additional Options. Set both together, with a store_id from Get Stores on the same domain, to get that one store’s assortment, price and availability. The store used is echoed in data.location. Store-targeted calls cost 2 credits and take 10-60 seconds; the node allows them up to 120 seconds. walmart.com and walmart.ca only.
Get Product accepts Domain (Walmart.com or Walmart.ca). Walmart.ca is only available together with Store ID and Delivery ZIP.
Device remains retired.

What changed in 0.13.0

Twenty-two new resources, 91 new operations. eBay, Target, Home Depot, Threads, Kuaishou, Zillow, Redfin, Booking.com, Airbnb, Tripadvisor, Yelp, Indeed, Glassdoor, Apple App Store, Google Play, G2, Capterra, Google Ads Transparency, Meta Ad Library, SEC EDGAR, Companies House and Extract are all new in this release. Nothing outside Walmart changed on the existing resources.

Walmart went from 2 operations to 7 and its parameters changed. Read the breaking-change block below before upgrading a workflow that uses Walmart.

Extract is not a platform. It takes any URL and returns raw HTML, readability Markdown, or plain text. Mode sets the price: Normal and Advanced are 1 credit, Ultra is 2. Nothing is billed unless the extraction succeeds, so a dead link, a bot wall, or a timeout is free.

Walmart changed in 0.13.0 (breaking)

Walmart was rebuilt and gained five operations. Existing Walmart nodes keep running, but re-check them — three of these changes are silent and two return a 400.

Delivery ZIP, Device and Store ID are gone, silently. Walmart retired them and they are no longer node parameters, so a saved value has nothing to send it and is dropped. A workflow that priced results against a ZIP or a store now gets Walmart’s default location back, with no error to tell you. There is no replacement: the response reports the location it actually used.
Start Page is now Page, and this is the silent one that costs you data. The old start_page field no longer exists on the node, so a saved value is never sent and every paginated Walmart node quietly resets to page 1. Re-open each Walmart Search node, add Page under Additional Options, and set it again.
Domain is now an enum and a saved value will 400. It used to be free text defaulting to walmart.com. It is now Walmart.com (United States) / Walmart.ca (Canada) / Walmart.com.mx (Mexico), sent as com, ca and com.mx. A node still holding walmart.com fails validation. It is also the price-bearing option: walmart.com.mx costs 2 credits, walmart.com and walmart.ca cost 1. It is offered on Search Products and Get Category Products only — those are the only two operations that accept it, and the only two that can cost 2.
Fulfillment Speed lost Anytime and 2 Days, and a saved value will 400. Only Today and Tomorrow remain. 2_days leaked 3-to-4 day items and anytime was a no-op — and anytime was the old default, so any node where you added this option without changing it now sends a value the API rejects. Remove the option instead of setting it to “anytime”.
Domain, Delivery ZIP, Device and Store ID are gone from Get Product entirely. Get Product takes Product ID and nothing else. Walmart.ca product pages could not be fetched at all, which is why Domain is search-and-category-only.
Sort By gained values and lost none. Rating: High to Low and New are new; the four you already had still work. This one is safe.
Min Price / Max Price are unchanged but are now in the currency of the selected domain, not always USD.
Five new operations: Get Reviews, Get Category Products, Get Buy Box Offer, Get Seller and Get Seller Products.
Get Buy Box Offer returns the winning offer only, never the full offer list.
Seller ID must be the numeric catalog seller ID (sellercatalogid on products and offers). The GUID form of the seller identifier returns 404.
Get Seller Products does not paginate. Roughly the first 40 server-rendered items come back; total_count reports the seller’s real catalog size.

Per-resource notes

Read these before you build. They are measured behaviours, not guesses.

Start with a lookup

Several resources are keyed by identifiers that only exist inside the target site’s own URLs. Each has a lookup operation — start there, then chain the ID it returns.

Kuaishou has the same shape for share links: Resolve Share Link turns a Kuaishou share URL into the user ID every other Kuaishou operation takes.

Pagination

Do not assume an operation pages. Several deliberately do not, and the node only exposes a paging field where one really exists.

Page number: Walmart Search / Reviews / Category, eBay Search, Target Search / Category, Home Depot Search / Reviews, Zillow Search, Redfin Search, Booking Search, Airbnb Search / Reviews, Tripadvisor Search / Reviews, Yelp Search / Reviews, Indeed Search / Company Reviews, Glassdoor Salaries, App Store Reviews, SEC Filings / Search, Companies House Search / Officers / Filing History, G2 Search / Reviews, Capterra Reviews, Trustpilot Search / Reviews / Category Businesses, Realtor.com Search / Agents, Rightmove Search / Sold / Agents, Mercado Libre Search / Reviews.
Cursor: Threads User Posts / User Replies / Post Comments, all cursor-paginated Kuaishou operations, Airbnb Search, Google Play Reviews, Google Ads Search, Meta Ads Search / Advertiser Ads. Mercado Libre Get Questions pages by Offset (next_offset from the previous page).
No pagination at all: every detail operation, plus App Store Search (raise Limit, up to 200 — every offset spelling is ignored), Google Play Search (one shelf of about 30 apps), Capterra Search (fixed at 20 results; ?page=2 returns identical rows), Walmart Get Seller Products (about 40 items), and Glassdoor Get Reviews (see below).
Airbnb Search takes either Page or Pagination Cursor, never both — a request carrying both is rejected. The cursor wins.
Google Play’s reviews cursor is opaque, single-use, and encodes the sort. Send it back with the same Sort it came from.
Meta Ads pages are 30 ads on page 1, then 10 per page. Each page is a credit, so depth costs roughly 10 ads per credit after the first page.

Slow, gated, or expensive surfaces

Glassdoor is slow and flaky. Per-attempt success was measured at 48% and per-call at 87%. A call typically takes 40 to 90 seconds; a failing one can run about 3 minutes before it answers 502. Raise the node timeout and retry on 502. Get Reviews is capped at three reviews per response by Glassdoor’s login wall — there is deliberately no page field. Move the window with Category and Employment Status, and read filteredreviewcount to see how many match. Addressing Get Reviews or Get Salaries by Employer ID needs an extra lookup; passing the reviewsurl / salariesurl that Get Company Profile returns is the single-fetch path.
G2 is 5 credits, the highest on the node, and a blocked page bills you. A blocked G2 page arrives as a billed 502. Retries are deliberately conservative. Get Product carries no review text — call Get Reviews for that.
Target is slow, not expensive. Product about 4s, search about 9s, category about 37s, reviews about 40s, and a 502-then-retry was seen at 105s. Budget timeouts accordingly. Get Reviews returns 8 review bodies maximum regardless of reviewcount; Max Review Bodies only trims. seller* is null for first-party stock, which is most of Target — null means “sold by Target”, not “unknown”.
Yelp Get Reviews page 1 is redundant. It re-fetches the document Get Business already returned and costs another 2 credits. Start at page 2.
Capterra and Tripadvisor page 1 of reviews are already inside the detail call. Use Get Reviews to page past it.
Trustpilot Get Reviews stops at page 10, 200 reviews per filter combination. Different stars x language x date range combinations are separate sets, so slice to go deeper. Get Business Profile already carries the 20 newest reviews in its Language, so skip Get Reviews page 1 when that is all you need.
Realtor.com Search stops at 10,000 results per query and Rightmove stops at 42 search pages, 40 sold pages and 50 agent pages. Slice by price band, property type or a smaller area to reach more.
Mercado Libre Search stops at 42 pages of about 48, about 2,000 results per query. Slice by price range, condition or seller. Get Product is the slowest call, up to about 10 seconds.
Home Depot search page size is fixed at 12 and reviews are 30 per page; asking past total_pages is a 404.

Silent-wrong-answer traps

These sites answer 200 with the wrong data rather than erroring, so the node closes the enum where it can. Where a value is a passthrough, treat an unexpected result as unfiltered rather than empty.

Zillow: a bare ZIP works alone but cannot be combined with a filter or a sort — Zillow geolocates that request shape and answers about another city. Use the city name there. On Listing Status = For Rent, Min Price / Max Price mean monthly rent.
Realtor.com: on Sold, soldprice is null in US non-disclosure states (Texas and the others listed under 0.19.0) while solddate is set. A null price there means undisclosed, not an error.
Mercado Libre: a query can be narrowed to a category, brand or line by Mercado Libre itself, and Brazil and Argentina may apply a “new” condition on their own. Check autoappliedfilters before treating a result set as complete. Units sold are ranges (“+500”), never exact counts.
Rightmove: there is no keyword filter, so the node offers none. Get Property on a live listing carries no price history; use Get Sold Property for an address’s past sales.
Redfin: city names are not accepted on Location. Pass a redfin.com region URL (/city/, /neighborhood/, /county/, /zipcode/) or Region ID + Region Type together. daysonmarket is always null on responses — do not build on it.
Booking.com: Check-In Date and Check-Out Date must be sent together, or Booking prices its own default range and returns real prices for dates you never asked for. A search with neither destination nor destination ID returns Booking’s homepage, costs a credit, and returns nothing.
Airbnb: a dateless search defaults to +30 days / 5 nights and A/B tests both the window and the prices — the response flags this as datesaredefaulted. Get Listing has no price field; nightly rates are search-only.
Indeed: Radius in Miles and Max Age in Days are closed sets because Indeed ignores anything else and returns the unfiltered set. Min Salary filters on Indeed’s own estimate for the role, so postings that publish no salary still match.
eBay: Results Per Page accepts only 60, 120 or 240 — eBay silently falls back to 60 for anything else. Get Seller Profile is a profile, not a catalog; to list a seller’s inventory use Search Listings with Search By = Seller, which works with no keyword. Turning on Sold Listings searches completed sales, and eBay publishes no headline count there, so total_results is null.
App Store: Country decides price, currency, localised title, and whether the app is sold there at all. Get Reviews is numeric App ID only (Get App also accepts a bundle ID), cannot 404 — an unknown ID and a real app with zero reviews return the same empty feed — and hard-stops at page 10, which is 500 reviews per storefront.
Meta Ad Library: totalresults caps at 50000 with totalis_capped: true; Meta only reports “>50,000”. Spend, reach, impressions and the paid-for-by disclosure are null on commercial ads — set Ad Type to political and issue ads to see them.
Google Ads Transparency: impressions and reach are EEA-only, because that is where disclosure is legally compelled. US creatives return null there. The three ad-format sets are disjoint — an advertiser’s text, image and video ads share no creatives. Totals come back as totaladsmin / totaladsmax; Google publishes a range, never an exact figure.
SEC EDGAR: XBRL concept tags are case-sensitive (netincomeloss is a 404, NetIncomeLoss is a match) — use Get XBRL Facts Index to list what a filer actually reports. Full-text search coverage starts in 2001.
Companies House: search is capped at page 50 — the register serves a 1000-result window per term whatever hit count it prints. Officers and filing history have no upper bound and answer an empty list past the last page, which is indistinguishable from a company with none.
Kuaishou reports its own failures inside HTTP 200 bodies, which the API surfaces as a 502. A 502 on Get Video usually means a bad identifier, not an outage.
Threads Get Post Comments takes a post ID only, never a handle. Address profiles and user feeds by user ID where you can: username lookups cost 4 credits instead of 2 because a handle needs a second lookup.

What changed in 0.12.0

TikTok pagination works again. Cursor was sent as a number on every paginated TikTok operation and the API rejected it with a 400. It is a string now. If you built an expression that fed a number into it, wrap it in String().
TikTok Sort By lost its invented values. User Posts is Latest / Popular and Search Videos is Relevance / Most Liked; the extra values never existed on TikTok and the old labels were wrong. A node saved with one of the removed values now falls back to 0.
Eleven Google operations added: Maps Place Details, Maps Reviews, Shopping Search, Shopping Product, Shopping Product Stores, Flights Search, Hotels Search, Hotels Detail, News Search, Trends and Trending Now. Existing Search, AI Mode and Maps Search nodes are untouched.
Walmart gained its real filters: Start Page (the only Walmart pagination field), Domain, Device, Fulfillment Type, Delivery ZIP and Store ID, on Search and, where the API supports them, on Get Product. Most of these were retired again in 0.13.0 — see the breaking-change block above.
Reddit Get Post takes a post ID. Pick Post ID or Post URL under Lookup By. It returns the post only; comments come from Get Post Comments.
Instagram follower and following lists accept up to 100 per page, matching the API. Post, reel and tagged feeds stay capped at 50.

Amazon changed in 0.10.0 (breaking)

The Amazon API now returns a clean normalized shape instead of a raw passthrough, and the node changed with it. Existing Amazon nodes keep running, but re-check them:

Domain is gone, replaced by Country. The API wants an ISO 3166-1 alpha-2 marketplace code, not an amazon. string. Two do not match their suffix: amazon.com is us and amazon.co.uk is gb. A saved Domain value is no longer sent, so a node that was set to de now returns the US marketplace until you set Country to Germany.
Sort By is gone. The marketplace accepts every sort value and ignores all of them — price-asc-rank, price-desc-rank, review-rank and date-desc-rank all return the identical unordered set. It was a filter that did nothing, so it is no longer offered.
Start Page is now Page, under Additional Options. One page per call, 1 credit per call.
Min Price and Max Price are gone. They were never in the API request schema and the backend dropped them silently. Filter on price in a downstream node instead.
Response fields are renamed. Products now expose asin, title, url, image, price, currency, rating, reviewscount, issponsored, position, badge, salesvolume and delivery. Expressions reading the old provider keys (urlimage, productname, totalreviews, isbestseller, buybox) need updating.
Get Offers is new. All seller offers for an ASIN — price, seller, condition, shipping, discount and which offer holds the buy box. 1 credit. Page 1 only: the response reports hasmorepages, but there is no verified way to request a later page.

TikTok Shop

Two things to know about TikTok Shop before you build on it:

Get Product returns no price. TikTok masks the price on the product page. Exact prices come from Search Products, Get Shop Products and Get Category Products.
Search Products -> Get Product is not a reliable pipeline. Only about 44% of the product IDs returned by Search Products resolve on Get Product, because TikTok has no detail data for the rest. Those IDs answer with HTTP status 404 and this body:

  { "error": "Product not found in this region.", "creditsused": 1, "creditsremaining": 997 }

There is no data key on that response, so branch on the status code — never on a data field. A 404 here is a normal outcome, not an error: skip the item rather than retrying it.

Get Product Reviews still works for many IDs Get Product cannot resolve. Measured on 8 IDs that 404 on detail, 8 of 8 returned HTTP 200 from Get Product Reviews and 7 of 8 carried at least one review. A measured sample rather than a guarantee, but reviews are worth one call before dropping the product.

Templates

Ready-to-use workflow templates built on this node. Find them in the n8n template library (search “Scavio”), or import the JSON directly in n8n via Workflows -> Import from File, then attach your Scavio API credential.

| Template | What it does | Product |
| — | — | — |
| Google Keyword Rank Tracker | Daily organic rank tracking for your keywords into Google Sheets | source=n8n-node&utmmedium=package&utm_campaign=readme”>YouTube |
| Amazon Price-Drop Alerts | Email alerts when a watched Amazon product hits your target price | source=n8n-node&utmmedium=package&utm_campaign=readme”>Walmart |
| Reddit Brand Monitor | Daily email digest of new Reddit mentions | source=n8n-node&utmmedium=package&utm_campaign=readme”>Instagram |
| TikTok Competitor Monitor | Alert and log when tracked TikTok accounts post | source=n8n-node&utmmedium=package&utm_campaign=readme”>TikTok |
| TikTok Hashtag Trends | Log trending videos per hashtag | source=n8n-node&utmmedium=package&utm_campaign=readme”>Instagram |
| Instagram Top Posts Ranker | Rank a handle’s recent posts by engagement | source=n8n-node&utmmedium=package&utm_campaign=readme”>Instagram |
| Instagram New-Post Alerts | Email the moment a watched handle posts | source=n8n-node&utmmedium=package&utm_campaign=readme”>Instagram |
| TikTok Engagement Rate Checker | Weekly engagement rate per handle into Google Sheets | source=n8n-node&utmmedium=package&utm_campaign=readme”>TikTok |
| TikTok Bio and Link Change Alerts | Slack alert when a creator edits their bio or link | source=n8n-node&utmmedium=package&utm_campaign=readme”>TikTok |
| TikTok New-Video Alerts | Email the moment a watched handle posts a video | source=n8n-node&utmmedium=package&utm_campaign=readme”>Google Maps |
| Google AI Overview Citation Tracker | Slack alert when your brand gains or loses an AI Overview citation | source=n8n-node&utmmedium=package&utm_campaign=readme”>dashboard.scavio.dev/sign-up — 50 credits to start.

Credentials

You need a Scavio API key. Sign up at source=n8n-node&utmmedium=package&utm_campaign=readme”>Scavio docs

source=n8n-node&utmmedium=package&utm_campaign=readme”>Compare Scavio vs alternatives
n8n community nodes documentation

About Scavio

source=n8n-node&utmmedium=package&utmcampaign=readme”>search API for AI agents — one API key, structured JSON. A real-time source=n8n-node&utmmedium=package&utmcampaign=readme”>SerpAPI alternative with data from:

source=n8n-node&utmmedium=package&utmcampaign=readme”>Amazon Product API and source=n8n-node&utmmedium=package&utm_campaign=readme”>Mercado Libre (Mexico, Brazil, Argentina, Colombia) — listings, product detail, reviews, buyer questions, and best sellers
eBay, Target, Home Depot and source=n8n-node&utmmedium=package&utmcampaign=readme”>YouTube API, source=n8n-node&utmmedium=package&utm_campaign=readme”>Instagram API — video and social media data
source=n8n-node&utmmedium=package&utmcampaign=readme”>X API and source=n8n-node&utmmedium=package&utmcampaign=readme”>Realtor.com and source=n8n-node&utmmedium=package&utm_campaign=readme”>Trustpilot — business profiles, TrustScores, filtered reviews, and category rankings
Google Ads Transparency and Meta Ad Library — live competitor ad creatives
SEC EDGAR and Companies House — filings, XBRL financials, officers, and company registers
source=n8n-node&utmmedium=package&utmcampaign=readme”>API key and explore the

Install

npm install n8n-nodes-scavio

This Week's Favorites in E-commerce

WhatsApp via QR (OnlineLiveSupport) 880/wk

Asaas 762/wk

OpenPix 683/wk

More in E-commerce

Hotmart 522/wk

NetSuite 210/wk

EasyShopper 53/wk

Kiwify 34/wk

Easyparser 98/wk

Hotmart 535/wk

TracePass 337/wk

Authentication 153/wk

Collmex 19/wk

Popular Tutorials

The Easiest Way to Build Your First Automation (n8n Tutorial)

From Zero to Your First AI Agent in 25 Minutes (No Coding)

n8n Will Change Your Life as a Developer

I Built a Marketing Team with 1 AI Agent

An AI Agent That Builds Teams of AI Agents