ebay-sold-listings-search skill

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}&LHSold=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 (LHPrefLoc=98) · domestic (LHPrefLoc=1, US Only on ebay.com) · worldwide (LH_PrefLoc=2)
• --itemCondition: any (no filter) · new (LHItemCondition=1000) · used (LHItemCondition=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 LHComplete=1 — Best-Offer-Accepted detection works) or false (BOA appears as plain buyit_now).
• --ipg: items per page — 60 (default), 120, or 240. Larger ipg fetches more per page but pages get heavier.
• --page: page number for pagination (_pgn), default 1.

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",                                       // final displayed price (string, 2-decimal). For BOA, this is the asking price — actual accepted offer is not disclosed by eBay
  "soldCurrency": "USD",         

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", "label": "Jewelry & Watches", "depth": 0 }
  ]
}

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):

a. Initialize page=1, collected=[], seenIds=set() b. Loop while len(collected) < count:

1. For each {keyword} in the keyword list:

c. Persist collected items keyed by keyword (write JSONL to disk to survive interruptions).

• 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.

1. 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