ebay-sold-listings-search

name

ebay-sold-listings-search

description

eBay sold-listings scraper across 8 marketplaces (ebay.com/.co.uk/.de/.fr/.it/.es/.ca/.com.au). Takes keyword plus filters (category, price range, item condition, item location, sort order, completed toggle) and returns paginated real-sale records with itemId, url, title, condition, conditionId, endedAt, soldPrice, soldCurrency, listingType (best_offer_accepted / buy_it_now / auction), isBestOfferAccepted, buyingFormat, bidCount, shipping, totalPrice, thumbnails, seller info, sellerType. Use when user mentions ebay sold, ebay sold listings, ebay sold prices, ebay completed listings, ebay comps, ebay resale prices, ebay auction sold results, ebay best offer accepted, scrape ebay sold, ebay market research, ebay pricing intelligence, ebay flipping research, real sold prices ebay, get sold prices from ebay. Also applies to price benchmarking, sold-price analytics, resale valuation, cross-marketplace price arbitrage, appraisal for collectibles, brand demand tracking on ebay.

eBay — Sold Listings Search

Keyword + filters → paginated list of items that actually sold on eBay, with final price, sold date, condition, listing format, shipping, seller, and images.

Language

All process output to user (progress updates, process notifications) follows the user's language.

Objective

Given one or more search keywords, fetch real sold (completed) listings from a chosen eBay marketplace with a user-defined filter set (category, price range, item condition, item location, sort order, completed-listings toggle), paginate up to the desired result count per keyword, and return one structured record per confirmed sale.

Prerequisites

• Target eBay marketplace search page reachable in the browser: e.g. https://www.ebay.com/sch/i.html?_nkw={keyword}&LH_Sold=1. No login required.
• Ability to run python, browser-act, and bash (for eval "$(python ...)").

Pre-execution Checks

1. Tool Readiness

If browser-act has been confirmed available in the current session → skip this step.

Invoke browser-act via Skill tool to load usage. If installation or configuration issues arise, follow its guidance to resolve then retry.

Capability Components

This Skill's operational boundary = what the user can manually do in their browser. It only reads data already displayed to the user on the search results page, never bypassing authentication or access controls. Its role is equivalent to copy-pasting on the user's behalf — the sold data is already on screen, automation merely saves time. JS code is encapsulated in Python files under the scripts/ directory, invoked via eval "$(python scripts/xxx.py {params})". $(...) is bash syntax; use the bash tool for execution.

Below are all atomic capabilities discovered and verified during exploration. Combine freely as needed.

URL Builder: assemble a sold-listings search URL for any of the 8 marketplaces

URL=$(python scripts/build-url.py '{keyword}' --ebaySite {ebaySite} --sortOrder {sortOrder} --categoryId {categoryId} --subcategoryId {subcategoryId} --itemLocation {itemLocation} --itemCondition {itemCondition} --minPrice {minPrice} --maxPrice {maxPrice} --includeCompletedListings {true|false} --ipg {60|120|240} --page {pageNumber})

The script prints one URL to stdout — capture into a bash variable and pass to navigate.

Parameters:

• {keyword} (positional, required): search term. 1–6 terms should each be run through a separate call.
• --ebaySite: ebay.com (default) · ebay.co.uk · ebay.de · ebay.fr · ebay.it · ebay.es · ebay.ca · ebay.com.au
• --sortOrder: endedRecently (default, _sop=13) · timeNewlyListed (_sop=10) · pricePlusPostageLowest (_sop=15) · pricePlusPostageHighest (_sop=16) · distanceNearest (_sop=7)
• --categoryId: main category filter (_sacat); default 0 (All Categories). Common IDs: see "Enum Parameters" section.
• --subcategoryId: subcategory ID; when set, overrides --categoryId.
• --itemLocation: default (LH_PrefLoc=98) · domestic (LH_PrefLoc=1, US Only on ebay.com) · worldwide (LH_PrefLoc=2)
• --itemCondition: any (no filter) · new (LH_ItemCondition=1000) · used (LH_ItemCondition=3000) · or pass any numeric eBay condition id (e.g. 1500 for Open Box, 2010 for Excellent Refurbished, 7000 for Parts). See enum section.
• --minPrice / --maxPrice: numeric filters _udlo / _udhi.
• --includeCompletedListings: true (default, adds LH_Complete=1 — Best-Offer-Accepted detection works) or false (BOA appears as plain buy_it_now).
• --ipg: items per page — 60 (default), 120, or 240. Larger fetches more per page but pages get heavier.

Output: a single fully-qualified URL string on stdout.

DOM: extract sold-listing cards from the current eBay search results page

Prerequisite: browser is currently on an eBay sold-listings search URL and the results have rendered.

1. navigate "$URL" (using URL from build-url.py)
2. wait --selector "li.s-card" --state visible --timeout 30000 (eBay may show a splash challenge before the results page; the splash resolves itself and li.s-card appears)
3. eval "$(python scripts/extract-page.py --keyword '{keyword}')"

Parameters of extract-page.py:

• --keyword: pass-through tag written to each item's keyword field so multi-keyword batches can be regrouped later. Optional; empty by default.

The extractor filters out eBay's promotional "Shop on eBay" placeholder cards automatically, only keeping cards with a real data-listingid.

Output example (top-level shape):

{
  "error": false,
  "pageUrl": "https://www.ebay.com/sch/i.html?_nkw=iphone+15&LH_Sold=1&LH_Complete=1&_ipg=60&_pgn=1&_sop=13",
  "heading": "17,000+ results for iphone 15",   // total count string as shown on the page; may say "10,000+" for very large sets
  "categoryId": null,                            // _sacat from URL, null if All Categories
  "category": "Cell Phones & Smartphones",       // label from the eBay category dropdown
  "itemCount": 60,                               // number of real items (excludes promo cards)
  "hasNextPage": true,                           // whether a "Next" pagination link exists
  "nextUrl": "https://www.ebay.com/sch/...&_pgn=2", // href of the Next link, if present
  "items": [ /* array of item records */ ]
}

Per-item record example:

{
  "itemId": "198484429340",                                   // eBay item ID
  "url": "https://www.ebay.com/itm/198484429340",             // canonical item URL (query stripped)
  "title": "Apple iPhone 15 Pro 128GB Blue A2848 Triple ...", // localized card title
  "condition": "Pre-Owned",                                    // localized condition label as displayed (Pre-Owned, Gebraucht, Neuf, Nuovo, Usado, Open Box, etc.); null when not shown
  "conditionId": 3000,                                         // best-effort numeric eBay condition ID; null when the label isn't in the lookup table
  "categoryId": null,                                          // effective _sacat used for the search; null when All Categories
  "category": "Cell Phones & Smartphones",                     // category label from the search page dropdown
  "endedAt": "2026-07-08T00:00:00.000Z",                       // sale completion date (parsed from "Sold X" caption, midnight UTC)
  "soldPrice": "437.40"                                       
                                          
                           
                                    
                                     
                                               
                                        
                                      
                                         
                                         
     
    
                                 
                                   
                                      
                                             
                                          
                          

Error handling: when the extractor returns {"error": true, ...}, check window.location.href and document.title via state — the page may be an eBay splash challenge (ebay.com/splashui/challenge), an anti-scrape "SORRY / Something went wrong" error page, or a redirect to a different page (very large ambiguous queries occasionally auto-redirect to a similar-item page). Wait longer or re-navigate; if repeatedly blocked, switch to a stealth browser with a different IP.

DOM: enumerate eBay top-level categories from the search page

Prerequisite: currently on any eBay search results page (any keyword).

eval "$(python scripts/enum-categories.py)"

Output example:

{
  "error": false,
  "count": 34,
  "items": [
    { "id": "20081", "label": "Antiques", "depth": 0 },
    { "id": "550", "label": "Art", "depth": 0 },
    { "id": "58058", "label": "Computers/Tablets & Networking", "depth": 0 },
    { "id": "281",     
  

depth: 0 = top-level main category. Subcategories with depth >= 1 may also appear when the current search is scoped to a category. For deep subcategory IDs beyond top-level, users should first search within the desired top category, then re-run the enum.

Composite: batch-collect sold listings across multiple keywords with pagination

For each keyword in the caller's list, loop pagination until either the caller-defined count (max items) is reached OR hasNextPage=false. Between pages, back off briefly (2–4 seconds) to avoid triggering rate limits. Deduplicate by itemId if a keyword sees the same item on adjacent pages.

Cross-page (contains navigate + non-JS steps):

1. For each {keyword} in the keyword list: a. Initialize page=1, collected=[], seenIds=set() b. Loop while len(collected) < count:
   • URL=$(python scripts/build-url.py '{keyword}' --ebaySite {ebaySite} --sortOrder {sortOrder} --categoryId {categoryId} --subcategoryId {subcategoryId} --itemLocation {itemLocation} --itemCondition {itemCondition} --minPrice {minPrice} --maxPrice {maxPrice} --includeCompletedListings {includeCompletedListings} --page $page)
   • navigate "$URL" → wait --selector "li.s-card" --state visible --timeout 30000
   • RESULT=$(browser-act --session {name} eval "$(python scripts/extract-page.py --keyword '{keyword}')")
   • Parse RESULT JSON. If error=true, retry once after a longer wait; if still failing, mark keyword incomplete and break.
   • For each item, if itemId not in seenIds: add to collected and seenIds.
   • If hasNextPage=false or len(collected) >= count: break.
   • sleep 2-4s, increment page. c. Persist collected items keyed by keyword (write JSONL to disk to survive interruptions).
2. Merge all keyword outputs into a single dataset.

Optional post-filter: if the caller supplied daysToScrape, discard items whose endedAt is older than now - daysToScrape days after collection (eBay's URL does not accept a direct sold-date range; sorting by endedRecently and stopping when items fall out of the window is the standard approach).

Enum Parameters

[DOM] categoryId / subcategoryId — retrieve top-level _sacat values via eval "$(python scripts/enum-categories.py)" (see "DOM: enumerate eBay top-level categories" above). Subcategories require first navigating to a category-scoped search. eBay publishes exhaustive category trees on its own site (footer link "See all categories") for deep hierarchies.

[Static] sortOrder — fixed mapping, no runtime enumeration needed: endedRecently=13, timeNewlyListed=10, pricePlusPostageLowest=15, pricePlusPostageHighest=16, distanceNearest=7. Applied by build-url.py.

[Static] itemLocation — fixed mapping: default=98, domestic=1, worldwide=2. Applied by build-url.py.

[Static] itemCondition — common eBay condition IDs (pass any numeric ID directly via --itemCondition):

• 1000 New / Brand New
• 1500 New Other / Open Box
• 1750 New with defects
• 2000 Certified Refurbished · 2010 Excellent - Refurbished · 2020 Very Good - Refurbished · 2030 Good - Refurbished
• 2500 Seller Refurbished
• 2750 Like New
• 3000 Used / Pre-Owned
• 7000 For parts or not working

[Static] ebaySite — fixed list of 8 marketplaces: ebay.com, ebay.co.uk, ebay.de, ebay.fr, ebay.it, ebay.es, ebay.ca, ebay.com.au. Applied by build-url.py.

Pagination

URL Pagination: URL pattern ...&_pgn={N} (1-based). Next page link selector: a[type=next], a[rel=next] — the extractor exposes it as nextUrl and hasNextPage in the response. Termination: hasNextPage=false, or items.length === 0, or endedAt of the last item is older than the caller's daysToScrape window when sorted by endedRecently, or the caller's count reached.

Success Criteria

error === false && itemCount >= 1 && items.every(i => i.itemId && i.url && i.title && (i.soldPrice !== null))

For pagination flows: unique itemIds across all collected pages >= min(count, hasNextPage-terminated total) and every item has soldPrice non-null.

Known Limitations

• eBay does not disclose the actual accepted amount for Best-Offer-Accepted sales. soldPrice for BOA items is the asking price at listing time, and isBestOfferAccepted=true signals the discrepancy. This is a data-source limitation, not a scraper limitation.
• BOA detection requires includeCompletedListings=true (the default). With LH_Complete=0, eBay strips the strikethrough price styling and "Best offer accepted" caption, so BOA sales appear as ordinary buy_it_now. All results are still genuinely sold either way (LH_Sold=1 is always applied).
• eBay's URL does not accept a direct sold-date range filter. Time-window scoping (e.g., "last 30 days") is achieved by sorting by endedRecently and stopping pagination when items fall out of the desired window.
• sellerType (private / business) is only labeled on EU marketplaces (ebay.de, ebay.fr, ebay.it, ebay.es); non-EU sites return null.
• fullResThumbnailUrl is derived by rewriting s-l500 → s-l1600 in the thumbnail URL. For items where eBay only hosts a low-res image, the URL may 404.
• IP-based rate limiting and splash challenges: aggressive pagination from a single IP can trigger ebay.com/splashui/challenge interstitials (typically resolve themselves within seconds) or a "SORRY / Something went wrong" error page (requires switching IPs or waiting). Space out requests 2–4 seconds between pages.
• Some marketplaces (ebay.co.uk, ebay.de) may refuse requests from IPs geolocated in unrelated regions with an outright error page. Use a stealth browser bound to a matching-region proxy (e.g., UK IP for ebay.co.uk) for reliable access to non-US sites.
• Localized condition labels are covered for English, German, French, Italian, Spanish. Novel or newly introduced eBay condition labels not present in the lookup table return conditionId: null while condition still shows the localized text.

Execution Efficiency

• Batch orchestration: For multi-keyword collection, loop keywords serially within a single browser session; only page-by-page pagination is safe from a single IP. Do not parallelize pages of the same keyword within one browser — the anti-scrape system tracks sessions.
• Higher throughput: For truly large batches, distribute keywords across multiple stealth browsers with independent fingerprints/proxies, one keyword per session. Each session has its own rate-limit budget.
• Test before batch execution: After wiring up the batch script, first run 1–2 keywords × 2 pages to verify parsing quality (spot-check listingType, soldPrice, endedAt), only then run the full batch.
• Reduce redundant pre-operations: Reuse the same browser session across keywords when possible; navigation between search URLs is cheap.
• Error resumption: Persist per-keyword results as JSONL to disk after each page; on failure, resume from the last completed page rather than restarting the keyword.
• Larger _ipg: Setting --ipg 240 (max) reduces the number of paginated requests four-fold, at the cost of heavier pages. Use for exhaustive collections; keep 60 for spot checks.

Experience Notes

Path: {working-directory}/browser-act-skill-forge-memories/ebay-sold-listings-search.memory.md (working directory is determined by the Agent running the Skill, typically the project root or current working directory)

Before execution: If the file exists, read it first — it records unexpected situations encountered during past executions (e.g., a strategy has become ineffective); adjust strategy order accordingly.

After execution: If an unexpected situation is encountered (strategy became ineffective, page redesigned, anti-scraping upgraded, better path discovered), append a line: {YYYY-MM-DD}: {what happened} → {conclusion}

Normal execution does not write to the file. Do not record what keywords were used or how many results were returned — those are task outputs, not experience.