Wildberries Search & Product Scraper avatar

Wildberries Search & Product Scraper

Pricing

from $0.10 / 1,000 listing rows

Go to Apify Store
Wildberries Search & Product Scraper

Wildberries Search & Product Scraper

Collect Wildberries search results, nmId product lookups and gallery images. Monitor product prices, ratings and review counts with CommerceStack; inspect explicit coverage and source errors.

Pricing

from $0.10 / 1,000 listing rows

Rating

0.0

(0)

Developer

Aleksandr Karavaev

Aleksandr Karavaev

Maintained by Community

Actor stats

0

Bookmarked

29

Total users

8

Monthly active users

9 days ago

Last modified

Categories

Features

Limited permissions

Share

CommerceStack collects public Wildberries search results, looks up products by article number, and retrieves product cards and gallery photos. The Actor starts a browser session, then collects the requested data over HTTP. It returns one Dataset row per unique nmId. It does not detect duplicate listings or compare images.

Use with ChatGPT, Claude and AI agents

This CommerceStack Actor is available through the official Apify MCP server. It is an Apify tool, not a separate built-in CommerceStack app in every AI assistant. The client must support remote MCP and be connected to your Apify account. Actor runs use the current Apify pricing.

  • Claude: connect the official Apify connector, then name commercestack/wildberries-search-product-scraper in your request.
  • ChatGPT: where your account and workspace allow custom MCP plugins, follow OpenAI's connection instructions. Use the server URL below and Apify OAuth.
  • Cursor, Claude Code, Codex and other MCP clients: add the same remote server using the client's MCP configuration. See Apify client setup. OAuth avoids pasting an API token into a conversation.

For this Actor plus its description, input schema and current pricing, use:

https://mcp.apify.com?tools=fetch-actor-details,commercestack/wildberries-search-product-scraper

Copy this first-run prompt:

Use commercestack/wildberries-search-product-scraper to collect up to 10 Wildberries search rows for "наматрасник 140х200" with search_only and one page. Read the schema and current pricing first. Return article numbers, titles and source price fields with units. Inspect MANIFEST and report incomplete coverage.

Before running, check the input schema, choose page/item limits and review the cost. Read the Dataset together with MANIFEST; an assistant should not infer a missing product, a final checkout price or complete coverage from an incomplete response. For repeated monitoring, keep the same watchlist key and delivery context and follow the monitoring instructions below.

Ready-to-use examples

Duplicate an example into your own Apify account and edit its input before running. Copying an example does not start it or enable a schedule.

For product links, use productUrls. This typed field works with the dedicated MCP tool; use numeric article IDs in products. Existing API requests with integer IDs, string IDs or canonical URLs in products remain supported. Both fields are combined and deduplicated by article number (200 total input entries maximum). Clear the example products value when using only your URLs.

{
"mode": "lookup_only",
"productUrls": ["https://www.wildberries.ru/catalog/192045310/detail.aspx"],
"maxItems": 1,
"timeoutSeconds": 60
}

Choose a mode

ModeInputDataset outputOpens product pages?
search_only1–5 search phrasesSearch-result rows, including article number, available seller and price fields, pics, rating, feedback count and search positionNo, after session startup
lookup_onlyUp to 200 article numbers or canonical product URLsThe same listing-format rows for articles found by WildberriesNo, after session startup
productsUp to 200 article numbers or canonical product URLsProduct-card fields and gallery image URLs or downloaded filesYes
search1–5 search phrasesProduct-card rows for results found in searchYes

search_only and lookup_only are the faster, lower-cost ways to collect listing data. They report the number of photos in pics when Wildberries provides it, but do not return photo URLs or image files. Use products or search for gallery images. Some fields depend on the data Wildberries supplies at collection time.

Search-result rows

{
"mode": "search_only",
"queries": ["наматрасник 140х200"],
"maxPagesPerQuery": 1,
"maxItems": 20,
"timeoutSeconds": 180
}

maxItems applies separately to each query in search_only. Each query can use up to two search pages. When the same product appears in several queries, it is returned once with all accepted query/page/position references.

Look up article numbers

{
"mode": "lookup_only",
"products": [192045310, "https://www.wildberries.ru/catalog/313247303/detail.aspx"],
"maxItems": 10,
"timeoutSeconds": 180
}

Missing articles are reported as product_not_found in the MANIFEST record. If all requested articles are missing and no technical error occurs, the run succeeds with MANIFEST.status: "partial", an empty Dataset and exit code 0. Enter article numbers in products, rather than as search phrases.

Product cards and photos

{
"mode": "products",
"products": [192045310],
"imageMode": "download",
"maxImagesPerProduct": 3,
"timeoutSeconds": 180
}

imageMode: "urls" returns observed gallery URLs without saving image files. imageMode: "download" saves up to six images per product in the run's key-value store and returns their URLs, MIME type, dimensions, byte count and SHA-256. Large full-card batches may reach the run deadline before collecting every requested product; maxItems is a limit, not a throughput guarantee.

Read the results

Dataset rows from search_only and lookup_only use schema wb-search-row-v1. Product-card rows use wb-products-v1. Both include nmId, canonical product URL, collection time and provenance. Listing rows can include name, brand, seller, stock, sizes, pics, matchId, subjectId, reviewRating and feedbacks. pics: 0 means Wildberries reported zero photos; null means it did not provide the field. Full-card rows can include descriptions, attributes, variants, price fields and an images array.

Price values are the original integers supplied by Wildberries, with their source field and units identified in the row. They are not a verified final RUB checkout price or a personal discount. Missing optional data remains null; availability and prices can change after collection.

The Actor's Output tab links to the Dataset and to MANIFEST in the default key-value store. Check MANIFEST.status, counts, errors, reachedLimits and, for search_only, coverage.perQuery. A platform run can finish successfully while a collection is partial. Rows already produced remain in the Dataset.

Pricing and spending limits

The Actor uses pay per event: $0.0001 per listing row (search_only or lookup_only), $0.0035 per product-card row (products or search), and $0.00625 per allocated GiB when a run starts. At the default 2 GiB, the start charge is $0.0125. For example, one listing row costs $0.0126 and 100 listing rows cost $0.0225 at 2 GiB. Platform usage is included in these prices; there is no separate charge for writing a Dataset item. Apify displays the applicable price and any Store discount before a run.

Set a maximum cost per run in Apify to bound event charges; the minimum accepted cap is $0.02. When that cap is reached, the Actor stops before collecting another product. Results already saved remain available and MANIFEST reports charge_limit and partial if only part of the requested batch was collected. The start charge applies even when no products are found. A larger memory setting increases the start charge and may leave fewer results within a small cap.

Limits and support

  • The supported region is RU-MOW (Moscow destination, RUB search context). Other regions and proxy settings are not supported in this version.
  • maxItems is at most 200; maxPagesPerQuery is at most 2; collection time is at most 600 seconds. The Actor rejects unsupported fields and malformed product URLs before starting the browser.
  • Browser access or Wildberries responses may change, and a request can return fewer items than expected. Review the manifest before relying on a batch.
  • Use the Issues tab on this Actor's Store page to report a reproducible problem. Include the run ID, input mode and affected nmId; do not post credentials or session cookies.

This independent Actor is not affiliated with or endorsed by Wildberries. Users are responsible for their use of the collected public data.

Switching modes in the Console

The selected mode determines which inputs are used. Search modes use queries and ignore a leftover products field. Product/lookup modes use products and ignore leftover queries and maxPagesPerQuery. Ignored fields are listed in MANIFEST.inputWarnings and in a startup warning; their values are not collected or charged. You must still provide the required inputs for the selected mode. Unsupported options and image downloads in listing-only modes remain errors.

Run diagnostics and sharing an error

The run status message summarizes the collection outcome, mode, saved Dataset rows and reason codes. Read MANIFEST in the default key-value store for the complete errors, limits and coverage. A successful platform run can contain a partial collection; confirmed missing products are different from technical failures. Status messages omit queries, product IDs, credentials and raw responses.

When reporting a problem, open an Issue on this Actor's Store page and include the run URL. Apify automatically shares that run with the developer, including its input, logs and storages. Share only runs whose contents you are comfortable sharing; never paste API tokens, passwords or cookies into an issue. Alternatively, you can enable Settings > Login & Privacy > Share run data with developers in Apify Console; this shares runs of public Actors with their developers more broadly.

Pinned builds keep their existing behavior. Test a new build with your normal input before updating a production pin. Historical failure counts alone do not identify a cause or prove a particular customer's run failed.

Start a daily price monitor

  1. Choose products, enable monitor, use imageMode: "urls", and start with the two-product example below. Set maxItems to cover the intended list; keep the price/rating/reviews fields unless you need another supported comparison.
  2. Choose Save as a new task. Keep the same task and monitorKey for each run. The key is a non-sensitive watchlist label, not an API credential. Separate keys are for independent watchlists, not retries of an uncertain charge.
  3. Run the saved task twice manually. The first successful observations are new; unchanged products in the second run receive compact rows. A real price, rating, review-count or variant change can still produce changed.
  4. Inspect the Dataset rows with schemaVersion: "wb-monitor-v1" and MANIFEST in the run's default key-value store. Check errors and delivered counts even when Apify reports SUCCEEDED. Deduplicate downstream delivery by eventId; a technical failure is not a product-removal event.
  5. In Schedules, select that task, set a daily time and timezone, and a maximum cost per run. Enable it only after the two manual checks. Keep resetMonitor: false; do not overlap runs sharing the key. Apify schedule setup.

For 100 products at the default 2 GiB, an illustrative event bill is $0.3625 for 100 first full-card observations and $0.0225 for 100 unchanged observations, including the start charge. One first run plus 29 unchanged daily runs is $1.015. This assumes all 100 observations are delivered, no discounts and no changed products or extra run starts; post-run storage is excluded. It is a price calculation, not a throughput guarantee. New/changed rows use the full-card price. Small batches still pay the start charge.

The Actor produces data in Apify; it does not automatically send email, chat messages or product-removal notices.

Scheduled product monitoring

Set monitor: true with mode: "products" and imageMode: "urls". Keep the same monitorKey (or saved task) for every scheduled run. History stays in named storage in your own Apify account. No third-party Actor is used.

{"mode":"products","products":[192045310,768543350],"maxItems":2,
"monitor":true,"monitorKey":"my-products","monitorFields":["price","rating","reviewsCount"],
"minPriceChangePct":3,"imageMode":"urls"}
  • First observation: changeType: new, full card and gallery URLs.
  • Relevant change: changed, full card with previous, changedFields and variantChanges.
  • No relevant change: unchanged, fresh short row; no expensive full-card visit.
  • Missing/blocked/incomplete response: MANIFEST error, no result charge and no baseline update.

Monitor rows use schemaVersion: wb-monitor-v1. monitorPrices maps WB optionId to sourceValue and a decimal RUB amount. This is the observed non-wallet product price (sizes[].price.product), not a personalized wallet or checkout quote. priceContext records seller, currency and requested destination -1257786. A verified delivery address is not claimed. Full rows retain raw prices for compatibility; use monitorPrices for normalized monitoring. observedAt is the listing observation; fullCardObservedAt is the separate enrichment time.

minPriceChangePct compares each variant with its last notified price. For 3%, 100 → 101 → 102 → 103 reports a price change at 103. A rating change does not reset that price baseline. A different seller/destination starts a new comparison context. Adding/removing variants reports variantSet. resetMonitor re-baselines successfully delivered observations; leave it false for normal scheduled runs.

No gone claim is made from an empty WB lookup. monitor_not_observed means the current response did not establish the product's state. Optional availability currently accepts positive stock only; zero or missing quantity is unknown and produces monitor_fields_unavailable. It is not a catalog-removal notification.

Monitoring price and limits

Existing event prices apply unchanged: one product-card for each new/changed row, or one listing-row for each unchanged row, never both for the same row. The usual Actor start charge also applies, including a new start when an interrupted run is resurrected. Errors and missing observations have no result event charge. Up to 200 IDs per run; maxItems and time limits still apply. New/changed products fetch full card data directly from WB's published CDN routes, with browser fallback if the route or response cannot be verified. Unchanged products need only the batched listing check. A lower result price does not mean every small batch has a lower total cost.

Delivery, recovery and notifications

Use an Apify scheduled task and the run-success webhook/integration to fetch the Dataset. Check MANIFEST even after a successful platform run: partial results are explicit. Consume eventId idempotently. No email or SKU data is sent outside Apify automatically.

Runs sharing a monitor key are serialized; monitor_busy means another run owns the lease. Use a later schedule or a separate key for independent monitoring. After interruption, resurrect the original run to reconcile its Dataset and confirmed charges. A new run cannot inspect another run's private billing data. monitor_reconciliation_required includes sourceRunId: do not reset or switch keys to replay the uncertain operation. Resurrect that run, or open an Issue with its URL so support can inspect the shared evidence. The Actor never automatically retries an uncertain charge. Exactly-once delivery across network failures is not promised; history advances only after confirmed delivery and billing.