eBay API Alternative: Items, Sellers and Feedback via API and MCP

If you need eBay listing, seller and feedback data inside a product, a spreadsheet or an agent, an eBay API alternative is a source that returns it as one JSON response per request. You send an item ID or a seller name over REST or MCP and get structured data back, with the point cost printed in the response. The eBay item source page is the reference for exact fields and prices.

The short version

• —What it is: pay-per-lookup sources that read public eBay listings, seller profiles and seller feedback at request time and return structured JSON. They read the marketplace from the outside; they do not act on your own seller account.
• —Who it is for: sourcing, pricing and risk teams, and agents that check a listing or a seller before a decision.
• —What it costs: 5 points for a listing or a seller profile, 1 point per returned row for feedback, seller listings and item search. Listed on the source pages as $5 per 1,000 and $1 per 1,000.
• —What it is not: a way to publish listings, answer buyers or manage orders, and not a source of ended or sold listings.

Where an eBay API alternative fits

The eBay sources return public marketplace information as data. They do not log in to anyone's account. It helps to separate three jobs that often get lumped together.

• —Reading the marketplace from the outside. What a listing says, what a seller's profile shows, what buyers wrote about that seller, what a store has for sale right now. This is the job these sources are built for.
• —Running your own eBay business through software. Publishing listings, updating stock or processing orders needs authenticated access to your own seller account. eBay offers its own developer programs for that; we have not verified their current terms, so check eBay's developer documentation. Nothing here replaces it.
• —Collecting the pages yourself. A scraper for pages you can see in a browser works until the markup changes, and then someone has to repair it. These sources exist so that the reading step is a request rather than a project.

The sources you will use

Four sources cover most questions a sourcing, pricing or risk team asks. The cost of each is the one stated on its source page.

Item search finds active listings by keyword, with filters for sort order, condition, buying format, price range and category, at 1 point per returned result. eBay categories returns the browse category tree, with the IDs you pass into item search, for 5 points flat. Field lists are on the seller feedback source page and its neighbours in the data source catalog.

Why the seller and the feedback are separate calls

A seller profile gives you the summary: how many feedback entries, what percentage is positive, how long the account has existed. The feedback source gives you the entries behind it. A high positive percentage can hide a pattern, such as late shipping or items not as described, and you only see that in the rows.

A seller check, step by step

The workflow is the same each time: look at the listing, then the seller, then the feedback, then the rest of the store. Every call uses the same shape, so you learn one request and reuse it.

1. Look up the listing

The input is one identifier: an item ID or a full listing URL in the item parameter. The source reads the public page at request time and turns it into fields. The price becomes a number, the currency a code, the images an array and the seller an object. Fields eBay does not publish for that listing come back as null rather than being guessed.

Here is the request for the eBay item source, taken from its source page:

curl -X POST https://api.neuralverge.ai/functions/v1/run-ebay-item \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "item": "188182070260" }'

The JSON below is the captured sample response published on that source page. It was captured for a different listing than the item ID in the example request, an Apple Magic Mouse, so do not expect the two to match. It is shown in full, and nothing in it has been added or edited by us.

{
  "session_id": "11111111-1111-1111-1111-111111111111",
  "kind": "ebay_item",
  "item": "336776720041",
  "human": "# eBay item\n\n- **Item:** 336776720041\n- **Proxy traffic:** 3.90 MB\n- **Title:** Apple Magic Mouse -A3204 Black - Wireless Bluetooth Mouse USB-C - Mint in Box\n- **Price:** USD75.99\n- **Condition:** Used\n- **Seller:** The Coco Store",
  "machine": {
    "item": {
      "id": "336776720041",
      "title": "Apple Magic Mouse -A3204 Black - Wireless Bluetooth Mouse USB-C - Mint in Box",
      "url": "https://www.ebay.com/itm/336776720041",
      "price": 75.99,
      "currency": "USD",
      "list_price": 79.99,
      "condition": "Used",
      "seller_notes": "“Excellent Mint Condition. Barely Used. Comes with Box.”",
      "quantity_available": null,
      "sold_count": null,
      "buying_formats": [
        "FIXED_PRICE"
      ],
      "has_best_offer": false,
      "image": "https://i.ebayimg.com/images/g/6rUAAeSw7BNqmbP2/s-l1600.png",
      "images": [
        "https://i.ebayimg.com/images/g/6rUAAeSw7BNqmbP2/s-l1600.png",
        "https://i.ebayimg.com/images/g/0VUAAeSw5NJqmbP~/s-l1600.png"
      ],
      "item_specifics": {
        "Condition": "Used",
        "Seller Notes": "“Excellent Mint Condition. Barely Used. Comes with Box.”",
        "Device Charging Range": "1-2",
        "Color": "Black",
        "MPN": "MXK63AM/A",
        "Tracking Method": "Laser",
        "Number of Buttons": "1",
        "Brand": "Apple",
        "Type": "Tactile/Multi-Touch Mouse",
        "Item Height": "0.85 in",
        "Maximum DPI": "3200",
        "Model": "Apple Magic Mouse",
        "Connectivity": "Wireless, Bluetooth, Lightning",
        "Charger Included": "No",
        "Features": "Rechargeable",
        "Country of Origin": "China"
      },
      "brand": "Apple",
      "mpn": "MXK63AM/A",
      "upc": null,
      "breadcrumbs": [
        "eBay",
        "Electronics"
      ],
      "category": "Mice, Trackballs & Touchpads",
      "seller": {
        "id": "cocosmartphones",
        "name": "The Coco Store",
        "feedback_count": 41808,
        "positive_percent": 99.8,
        "store_url": "https://www.ebay.com/str/cocosmartphones?_trksid=p4429486.m3561.l161210"
      },
      "watcher_count": 39
    },
    "proxy_traffic_bytes": 4086338
  },
  "total_points": 5
}

Every response carries a session_id you can trace, a kind naming the source, a short human summary, a machine object with the structured data, and total_points, the number of points that call cost. That last field is what lets you budget per call rather than per month. A few things in the JSON are worth pointing at:

• —total_points is 5. One successful lookup, five points, which the source page lists as $5 per 1,000 lookups.
• —price and list_price. The listing is at 75.99 USD against a struck-through 79.99. Both are numbers, so a discount calculation is arithmetic rather than string parsing.
• —condition and seller_notes. The structured condition is "Used", and the seller's own words are kept separately and are not verified by eBay or NeuralVerge.
• —item_specifics. Seller-entered key and value pairs. The source page says the shape varies per listing, so do not hard-code a list of keys.
• —Nulls. quantity_available, sold_count and upc are null in this capture. They are null because the listing did not publish them, not because something failed.

2. Look up the seller

The seller sources take a seller parameter: a username or a store URL. The profile request is { "seller": "cocosmartphones" } against run-ebay-seller, and it returns a seller object. We list the fields rather than the whole response to keep this article readable:

• —id, name and location
• —feedback_count and positive_percent
• —item_sold_count and follower_count
• —member_since, as eBay displays it
• —description, the storefront text

In the captured sample on that source page, the seller shows a feedback count of 35,588, 99.8 percent positive and a member-since date of Jan 20, 2008, for 5 points. New or low-volume sellers return sparse profiles.

3. Read the feedback behind the score

The request is run-ebay-seller-feedback with a seller and a max_items. max_items caps the number of rows (up to 500, default 25), and start_page pages beyond that. The sample on the source page was captured with max_items set to 2, and its total_points is 2, which is the 1-point-per-row rate in action. Here is one of its two rows, cut down to the fields that matter. Long comment text is truncated in the published sample itself:

{
  "id": "3024425293019",
  "rating": "positive",
  "from_user": "2***e",
  "from_user_score": 100,
  "role": "buyer",
  "item_id": "304930535351",
  "item_title": "In the Line of Duty IV Blu-ray (2023) Cynthia Khan, Woo-ping (DIR) cert 18",
  "item_url": "https://www.ebay.com/itm/304930535351",
  "price": 13.5,
  "currency": "GBP",
  "is_verified_purchase": true,
  "period": "Past month"
}

This is a subset of the row, with the text field left out for length. The buyer's username is partially masked by eBay, so do not plan on joining feedback to named people. Each row points at the item it relates to, so a cluster of complaints about one product or one kind of problem is something you can count.

4. Check what else the store sells

The seller items source returns a store's current active listings from the same seller input, at 1 point per returned row. Use it to compare a competitor's store with your own assortment. For keyword-led discovery instead of a known seller, use item search.

A worked example

Imagine Acme Oy, an illustrative Finnish company that resells refurbished peripherals. A buyer at Acme Oy spots a listing from a seller they have not used before. They look up the listing to confirm condition and price, look up the seller to see how long the account has existed and how many sales it has, then read a page of feedback and look for repeated complaints. If the rows show a pattern, Acme Oy skips the seller. If not, the purchase goes ahead.

REST and MCP

The same lookups are reachable two ways. Over REST it is the curl call above. Over MCP it is a tool call with the same argument. The eBay item, item search, seller, seller feedback and seller items sources are available as MCP tools named run_ebay_item, run_ebay_item_search, run_ebay_seller, run_ebay_seller_feedback and run_ebay_seller_items. Tool names use underscores; REST paths use hyphens.

{ "tool": "run_ebay_item", "arguments": { "item": "188182070260" } }

Use REST for backend jobs and scheduled price watching, and MCP for agents. The response is the same either way; parameters are in the documentation. For connecting a client to the hosted MCP server, follow the MCP server setup guide. For the wider picture of marketplace sources, see the e-commerce data API guide.

What it costs

Cost is printed in the response. The numbers below come from the source pages. We have not measured speed, hit rate or coverage, so we do not quote any.

A full due diligence pass on one seller, with the profile, 25 feedback rows and 25 current listings, is 55 points if every call returns what you asked for. On the stated rate of $1 per 1,000 points, that is $0.055. Current plan details, including how points are bought, are on the pricing page, which is the source to trust if it differs from this article. For the item and seller sources, the pages state that an expired item ID or an unknown username is not billed; for the row-based sources the bill is the rows actually returned, so a seller with no feedback costs nothing.

Limits and pitfalls

• —Active listings only. Item, item search and seller items cover listings that are live at request time. Ended auctions and sold buy-it-now listings do not resolve or drop out. If you need sold-price history, these sources are not that.
• —A snapshot, not a history. Price, watcher counts and feedback change between calls. If you need a trend, schedule the call and keep the results with your own timestamp.
• —Self-reported data. Feedback is left by buyers and the listing text is written by sellers. Neither is independently verified by eBay or by us, and the source pages say so.
• —Terms. Marketplaces have their own rules about how their data may be used. That check is yours to make. Keep the session_id with every stored record so a value can be traced back later.

How to decide

Run through these questions before you build on any marketplace data source, including this one.

1. —Is the job outside-in reading? If it needs your own seller account, it belongs with eBay's official programs, not here.
2. —Which workflow is it? The common ones map to one or two calls each:
   • —Competitor price watching: item search on your product keywords on a schedule, or the item source on specific competing listings.
   • —Supplier due diligence: the seller profile, then the feedback rows for patterns such as late shipping or items not as described.
   • —Sourcing for resale: item search filtered by condition and price, then a seller check before buying.
   • —Assortment gap analysis: a competing store's active listings against your own catalogue.
   • —Agent tool calls: the same sources as MCP tools.
3. —Do the fields you need exist? Anything eBay does not publish for a listing is null; compare the source page's field list with what you read.
4. —What is the cost per decision? Multiply the points by the calls a decision takes, as in the table above.
5. —Is a snapshot enough? If not, schedule the call and store the results.
6. —Have you checked eBay's own terms? That check is yours to make.

Frequently asked questions

What does an eBay item lookup cost?

The eBay item source costs 5 points per successful lookup, which its source page lists as $5 per 1,000. The eBay seller profile source is also 5 points. Item search, seller feedback and seller items cost 1 point per returned result, listed as $1 per 1,000. An item or seller that does not resolve is not billed.

Can I pass a listing URL instead of an item ID?

Yes. The item parameter accepts either an eBay item ID or a full listing URL. The seller sources work the same way and accept either a seller username or a store URL.

Does this replace eBay's official developer APIs?

Not necessarily, and this article does not compare feature lists because we cannot verify eBay's current terms. eBay's official programs may cover things these sources do not, such as managing your own listings or orders. Check eBay's developer documentation for eligibility, access rules and limits, then decide which fits the job.

Can an AI agent call the eBay sources through MCP?

Yes. The eBay item, item search, seller, seller feedback and seller items sources are available over REST and as MCP tools named run_ebay_item, run_ebay_item_search, run_ebay_seller, run_ebay_seller_feedback and run_ebay_seller_items.

Do ended or sold listings come back?

No. The item, item search and seller items sources only cover listings that are active at request time. An ended or removed listing does not resolve, and you are not billed for that request.

Is the data live or stored?

The source pages describe each lookup as read at request time. A price, a watcher count or a feedback score can change between two calls, so treat every response as a snapshot and add your own timestamp if you need a history.