---
title: "Best MCP Servers for Product and Price Data (2026)"
url: https://www.datawebot.com/blog/best-mcp-servers-product-price-data
description: "An MCP server lets an AI agent call a data tool directly. We compare 12 MCP servers for product and price data on tools, access, pricing and best use."
published: 2026-10-25
updated: 2026-10-25
author: Wilhendra Akmam
---

Blog

# Best MCP servers for product and price data in 2026

An MCP server lets an AI agent call a data tool directly. We compare 12 MCP servers for product and price data on tools, access, pricing and best use.

[Wilhendra Akmam](https://www.datawebot.com/authors/wilhendra-akmam) Published October 25, 2026

An MCP server for product data is a service that an AI agent, such as Claude, ChatGPT or a coding agent, can call directly to search products, read prices and list sellers, using the Model Context Protocol. This guide compares 12 of them, checked on each vendor's own pages on 25 October 2026.

MCP (Model Context Protocol) is an open standard for connecting agents to tools: the server publishes a list of tools with input schemas, and the agent decides when to call them ([spec](https://modelcontextprotocol.io/specification/2025-06-18/server/tools)). For product and price data, that means an agent can answer "who sells this, and for how much?" without you writing the integration. The servers below fall into three groups: shopping catalog servers that return products and offers, data-vendor servers for specific sites such as Amazon, and general web data servers where the agent brings the URL or the schema. One of the 12 is datawebot, our product, described in the same format, with a section on when not to use it.

## How different are seller prices for the same product?

This is the question most product data MCP servers exist to answer, so we measured it for this article.

On 25 October 2026 we asked datawebot's MCP server (country US) for six common products: Sony WH-1000XM5 headphones, Apple AirPods Pro (2nd generation), a Keurig K-Mini coffee maker, an Instant Pot Duo 7-in-1, a Ninja AF101 air fryer and a Kindle Paperwhite 16GB. For each, the agent ran `search_products`, picked the matching catalog product, then ran `get_product_offers` to list every seller.

| Measure | Result |
| --- | --- |
| Products checked | 6 |
| Seller offers returned | 57 |
| Median offers per product | 12 (range 1 to 13) |
| Median gap, most expensive to cheapest offer, all offers | 5.1 times (5 products with two or more offers) |
| Offers that were not a like-for-like match | 29 of 57 (51%) |
| Median gap after removing those 29 offers | 1.9 times (range 1.5 to 3.6 times) |

The 29 excluded offers were pre-owned or damaged units, used grades, bundles, older versions or a different voltage, out-of-stock listings, resale listings with no stated condition, unbranded titles, and a few links to unrelated websites. Even after removing them, the most expensive like-for-like seller asked a median of 1.9 times the cheapest, before shipping.

This is a small snapshot, not a market statistic: six products, one country, run between 12:12 and 12:13 UTC on one day. We planned 12 products; the account's free daily allowance ran out after six offer lookups, so we report the six we ran. The Kindle returned a single pre-order offer and is left out of the gap figures. Prices may have changed since. The run file (datawebot-seller-price-gap-2026-10-25.xlsx) lists each offer, its price, its shipping total where shown, and the reason for every exclusion; we can share it on request.

What it means for an agent: an offers tool gives a raw list, and the agent should check condition, model and seller before it calls one price "the cheapest". Tell your agent to do that in the prompt. Our guide to [product matching](https://www.datawebot.com/blog/product-matching) has the rules.

## How do the 12 MCP servers compare?

| MCP server | What it covers | Endpoint and sign-in | Pricing (as published) | Best for |
| --- | --- | --- | --- | --- |
| Shopify Global Catalog MCP | Products and offers across Shopify merchants | Remote, needs a UCP agent profile URL | Not stated | Comparing offers across Shopify merchants |
| Shop MCP (Shopify) | Shopify Catalog search plus checkout | Remote, search without credentials | Not stated | Agents that search and check out |
| ShopSavvy MCP | Offers across retailers, price history | Remote or local; anonymous or API key | Free without a key at 10 calls a minute; plans from $49 a month | Offers plus price history |
| BestPrice MCP | Greek shops via BestPrice.gr | Remote, no account | "Free for reasonable use" | Shopping in Greece |
| Canopy MCP | Amazon | Remote, OAuth 2.1 or API key | 100 free requests a month, then $0.01 a request | Amazon data from a data vendor |
| Bright Data MCP | Search, scrape and 60+ pre-built scrapers | Remote or local, API token | Free 5,000 requests a month | Many sites from one server |
| Oxylabs MCP | Amazon search and product, plus general tools | Remote or local, header credentials | Web Scraper API pricing; 1-week trial | Structured Amazon pages |
| Zyte MCP | Product extraction from any URL | Remote, sign-in through the client | MCP free; usage at Zyte API prices | Product fields from a known URL |
| SerpApi MCP | SerpApi search engines | Remote or self-hosted, API key | SerpApi plans; free 250 searches a month | Search pages, including shopping engines |
| Firecrawl MCP | General web scrape, search, crawl | Remote, OAuth or key; keyless mode | Free 1,000 credits a month; Hobby $19 | Schema-based extraction from any page |
| Apify MCP | Apify Store actors | Remote or local, OAuth or token | Apify plan usage plus each actor's price | Many site scrapers through one connection |
| datawebot | Product search across online shops; seller offers; Jobs for Amazon, Walmart and the stores in the [endpoint catalogue](https://www.datawebot.com/endpoints) | Remote, OAuth 2.1 or key; free email sign-up | Search and offers free within a daily allowance; Jobs from 200 free credits | Agents asking who sells a product and at what price |

## Which MCP servers return shopping catalogs and offers?

**Shopify Global Catalog MCP.** Shopify's [Global Catalog MCP](https://shopify.dev/docs/agents/catalog/global-catalog) has `search_catalog` (text, image and similar items) across "all Shopify merchants", `lookup_catalog` for up to 50 IDs or URLs, and `get_product`. Results are grouped by a Universal Product ID with offers from several merchants: price, list price, availability, seller and a checkout URL. Every request needs a UCP agent profile URL; no API key is mentioned. Pricing and rate limits: not stated on the page. Best for: comparing products and offers across Shopify merchants. Limitation: Shopify merchants only, and you must publish an agent profile.

**Shop MCP (Shopify).** Announced on 13 October 2026, [Shop MCP](https://shopify.dev/changelog/posts/add-shopping-and-shop-pay-checkout-to-any-agent-with-shop-mcp.md) adds checkout tools (`complete_checkout`, `get_checkout`) and `search_orders` to catalog search over "Over 1 billion products in Shopify Catalog". Search needs no credentials; saved details and Shop Pay need a Shop sign-in. Pricing: not stated. Best for: agents that search and also buy on Shopify stores. Limitation: Shopify catalog only.

Note that Shopify's older per-store Storefront MCP no longer serves this role: Shopify's docs say its catalog and cart tools "were removed" ([source](https://shopify.dev/docs/apps/build/storefront-mcp)). Older lists that recommend it are out of date.

**ShopSavvy MCP.** The [ShopSavvy MCP server](https://github.com/shopsavvy/shopsavvy-mcp-server) has 15 tools, including `product_search`, `product_lookup`, `product_offers` across retailers, `product_price_history` and scheduled monitoring (with a key). It runs remotely or locally with `npx`, under an MIT license. Pricing: free without a key at 10 calls a minute and 100 an hour; with a key, calls use plan credits, from $49 a month ([source](https://shopsavvy.com/data)). Best for: retailer offers plus price history inside an agent. Limitation: the anonymous limits are low, and the repo does not list which retailers or countries are covered.

**BestPrice MCP.** The [BestPrice.gr MCP](https://www.bestprice.gr/mcp) is read-only, with `search_products` (up to 8 results), `compare_offers` (shipping and delivered totals for a Greek postal code), `get_price_history` (daily minimum over 30, 90 or 180 days) and `get_shopping_decision`. No account or key is needed. Pricing: "Free for reasonable use", with rate limits and no numbers given. Best for: shopping in Greece, with price history. Limitation: Greece only, prices in euros.

**datawebot.** datawebot is a remote MCP server at `https://mcp.datawebot.com/mcp` (Streamable HTTP, OAuth 2.1 or a bearer key). Agents call `discover` first, then `search_products` to find products for sale across online shops (title, price, shop, rating, review count, image) and `get_product_offers` for every seller's offer on one product, with a direct shop link, shipping total and stock notes. Every result carries the time it was retrieved. The same server runs extraction Jobs for Amazon, Walmart and the stores in the [endpoint catalogue](https://www.datawebot.com/endpoints) through `list_job_blueprints`, `submit_job`, `get_job` and `get_results`.

Sign-up is free by email approval, with no card and no password. Search and offers are free within a daily allowance per account. Jobs use credits: 200 free trial credits per account, credit packs from $10 for 10,000 credits, and failed or empty Jobs are free ([pricing](https://www.datawebot.com/pricing)). To connect, tell a coding agent "set up [https://mcp.datawebot.com/setup.md](https://mcp.datawebot.com/setup.md)", or add the MCP URL as a custom connector in Claude (Settings, Connectors) or ChatGPT (Plus or above). Details are in [connect your agent](https://www.datawebot.com/docs/agents).

Best for: an agent that needs to answer "who sells this, and at what price?" across shops, with links. Limitation: results are best-effort snapshots; there is no stored price history, no scheduled checks and no alerts, and the free daily allowance is sized for lookups, not bulk runs.

## Which MCP servers come from Amazon data vendors?

**Canopy MCP.** [Canopy](https://www.canopyapi.co/) runs a hosted MCP server at `https://mcp.canopyapi.co/mcp` (Streamable HTTP, OAuth 2.1 or API key) covering Amazon product, pricing, Buy Box, offers, reviews, sales estimates and search. Pricing is the same as its API: 100 free requests a month, then $0.01 a request, or Premium from $99 a month. Best for: an Amazon-specific MCP from a data vendor. Limitation: Amazon only.

**Oxylabs MCP.** The [Oxylabs MCP](https://github.com/oxylabs/oxylabs-mcp) has 10 tools, including `amazon_search_scraper`, `amazon_product_scraper`, `google_search_scraper`, `universal_scraper` and AI Studio tools. It runs remotely with header credentials or locally with `uvx`. Pricing: the repo mentions a 1-week free trial of the Web Scraper API, and AI Studio comes with 1,000 free credits. Best for: structured Amazon search and product pages for agents. Limitation: the two Amazon tools are its only ecommerce tools, and with no OAuth yet, clients that only accept OAuth connectors need the local setup.

## Which general web data MCP servers can return product data?

These servers are not ecommerce-specific. They suit agents that already have the shop URLs, or that need a site no catalog covers.

**Bright Data MCP.** The [Bright Data MCP](https://github.com/brightdata/brightdata-mcp) offers search, markdown scraping, browser automation and 60+ pre-built scrapers found with `search_scrapers` and run with `run_scraper`, including Amazon and Walmart. It has 32 tools in groups ([docs](https://docs.brightdata.com/mcp-server/overview)). Pricing: free for 5,000 requests a month with no card, then $1.50 per 1,000 results for search, scrape and extract. Best for: covering many sites from one server. Limitation: product data goes through a scraper-discovery step, and many tools load by default unless you filter them.

**Zyte MCP.** The [Zyte MCP](https://docs.zyte.com/zyte-web-data/mcp.html) has fetch tools, `extract_from_http` and `extract_from_browser` (AI extraction of standard types including products, or custom fields described in plain language), `search`, and a domain pricing tool. Zyte says "Using Zyte MCP itself is free"; usage is billed at Zyte API prices, with $5 of trial credit for one month. Best for: product fields from a shop URL the agent already has. Limitation: there is no product catalog, and cost varies by site.

**SerpApi MCP.** The [SerpApi MCP](https://github.com/serpapi/serpapi-mcp) exposes one `search` tool over SerpApi's engines, with engine schemas published to the agent. It needs a SerpApi key; plans start with a free tier of 250 searches a month ([pricing](https://serpapi.com/pricing)). Best for: agents that already use SerpApi, including its shopping engines. Limitation: the README does not name the Google Shopping or Walmart engines, and results are search pages, not merged offers.

**Firecrawl MCP.** The [Firecrawl MCP](https://docs.firecrawl.dev/mcp-server) provides general web tools: scrape (with JSON or schema extraction), search, map, crawl, an agent tool and monitors, with a keyless mode for search and scrape within daily limits. Pricing: 1,000 free credits a month, Hobby at $19 a month for 5,000 and Standard at $99 for 100,000 ([pricing](https://www.firecrawl.dev/pricing)). Best for: pulling structured fields from arbitrary pages with a schema. Limitation: there are no ecommerce-specific tools, so you define the product schema and find the pages yourself.

**Apify MCP.** The [Apify MCP](https://docs.apify.com/platform/integrations/mcp) lets an agent search the Apify Store, read an actor's details and run it with `call-actor`, then read results with `get-dataset-items`. Runs use Apify plan usage (the Free plan includes $5 a month) and each actor's own price, for example "from $3.00 / 1,000 results" for an Amazon actor ([pricing](https://apify.com/pricing)). Best for: many different site scrapers through one connection. Limitation: a two-step flow, and output and quality vary by actor.

## When should you not use datawebot?

- **You need price history inside the agent.** ShopSavvy MCP and BestPrice MCP (for Greece) expose price history tools. datawebot returns snapshots; you would store them yourself, as in our [Python price tracker](https://www.datawebot.com/blog/price-tracker-python).
- **You need checkout.** Shop MCP is built for agents that buy. datawebot does not place orders; it returns links to the shop.
- **You need Shopify merchants' catalogs with checkout links.** Shopify's Global Catalog MCP returns offers grouped across Shopify merchants with checkout URLs.
- **You need deep Amazon fields.** Canopy MCP returns Buy Box, sales estimates and Amazon offers in depth.
- **You need bulk volume or an arbitrary site.** Bright Data, Apify, Zyte and Firecrawl are built for wide or custom collection.
- **You need alerts.** datawebot sends no alerts. ShopSavvy MCP lists scheduled monitoring with a key.

## How do you choose an MCP server for product data?

1. **Decide what the agent must answer.** "What is for sale?" needs search. "Who sells this exact item?" needs offers. "Was it cheaper last month?" needs history. Our guide to [price tracking APIs](https://www.datawebot.com/blog/price-tracker-api) explains the three.
2. **Check coverage on your own products.** Run five products you know well and compare the results to the shops' own pages.
3. **Check sign-in for your client.** Some chat clients accept only OAuth connectors; others need an API key in a config file. Check this before you build on a server.
4. **Check the free limits.** Anonymous and free tiers are useful for testing, but low per-minute or daily limits will stop a batch job, as our own snapshot run found.
5. **Tell the agent to clean offers.** In our snapshot, about half of the offers under one product were not like-for-like. A prompt line such as "only compare new, in-stock listings of the exact model" changes the answer.
6. **Look at the raw API too.** If you will call the data from code rather than an agent, compare the underlying APIs in [best ecommerce data APIs](https://www.datawebot.com/blog/best-ecommerce-data-apis).

## Frequently asked questions

### What is the best MCP server for product prices?

It depends on the job. For offers across shops with links, datawebot or ShopSavvy MCP. For Shopify merchants, Shopify's Global Catalog MCP. For Amazon, Canopy MCP or Oxylabs MCP. For any site you already have URLs for, Zyte, Firecrawl or Bright Data.

### Is there a free MCP server for price comparison?

Several have free use. ShopSavvy MCP works without a key at 10 calls a minute, BestPrice MCP is "free for reasonable use" in Greece, Canopy gives 100 free requests a month, Bright Data 5,000 a month, Firecrawl 1,000 credits a month, and datawebot gives free search and offers within a daily allowance. Shopify does not state pricing for its catalog MCP servers.

### Can Claude or ChatGPT compare prices across stores?

Yes, with an MCP server connected. Add the server's URL as a custom connector, then ask the agent to search and compare. For datawebot, use `https://mcp.datawebot.com/mcp`; ChatGPT needs Plus or above for custom servers. Steps are in [connect your agent](https://www.datawebot.com/docs/agents).

### Does Shopify have an MCP server for product data?

Yes. Shopify's Global Catalog MCP and Shop MCP search products across Shopify merchants, and Shop MCP adds checkout. The older Storefront MCP's catalog and cart tools were removed.

### How much do prices differ between sellers of the same product?

In our snapshot of six US products on 25 October 2026, the most expensive like-for-like seller asked a median of 1.9 times the cheapest. Across all 57 offers returned, before cleaning, the median gap was 5.1 times. It is a small sample, not a market average.

### Which MCP servers include price history?

ShopSavvy MCP has a price history tool, and BestPrice MCP returns daily minimums over 30, 90 or 180 days for Greek shops. datawebot returns dated snapshots only, and none of the other servers in this list describes a price history tool on the pages we checked.

## Sources

1. [https://modelcontextprotocol.io/specification/2025-06-18/server/tools](https://modelcontextprotocol.io/specification/2025-06-18/server/tools)
2. [https://shopify.dev/docs/agents/catalog/global-catalog](https://shopify.dev/docs/agents/catalog/global-catalog)
3. [https://shopify.dev/changelog/posts/add-shopping-and-shop-pay-checkout-to-any-agent-with-shop-mcp.md](https://shopify.dev/changelog/posts/add-shopping-and-shop-pay-checkout-to-any-agent-with-shop-mcp.md)
4. [https://shopify.dev/docs/apps/build/storefront-mcp](https://shopify.dev/docs/apps/build/storefront-mcp)
5. [https://github.com/shopsavvy/shopsavvy-mcp-server](https://github.com/shopsavvy/shopsavvy-mcp-server)
6. [https://shopsavvy.com/data](https://shopsavvy.com/data)
7. [https://www.bestprice.gr/mcp](https://www.bestprice.gr/mcp)
8. [https://www.canopyapi.co/](https://www.canopyapi.co/)
9. [https://github.com/brightdata/brightdata-mcp](https://github.com/brightdata/brightdata-mcp)
10. [https://docs.brightdata.com/mcp-server/overview](https://docs.brightdata.com/mcp-server/overview)
11. [https://github.com/oxylabs/oxylabs-mcp](https://github.com/oxylabs/oxylabs-mcp)
12. [https://docs.zyte.com/zyte-web-data/mcp.html](https://docs.zyte.com/zyte-web-data/mcp.html)
13. [https://github.com/serpapi/serpapi-mcp](https://github.com/serpapi/serpapi-mcp)
14. [https://serpapi.com/pricing](https://serpapi.com/pricing)
15. [https://docs.firecrawl.dev/mcp-server](https://docs.firecrawl.dev/mcp-server)
16. [https://www.firecrawl.dev/pricing](https://www.firecrawl.dev/pricing)
17. [https://docs.apify.com/platform/integrations/mcp](https://docs.apify.com/platform/integrations/mcp)
18. [https://apify.com/pricing](https://apify.com/pricing)
19. [https://www.datawebot.com/llms.txt](https://www.datawebot.com/llms.txt)
20. [https://mcp.datawebot.com/setup.md](https://mcp.datawebot.com/setup.md)

## Put product data in your agent's hands.

Connect a coding agent, or check how page-based credits work before you run anything.

[Connect your agent](https://www.datawebot.com/docs/agents) · [See pricing](https://www.datawebot.com/pricing)

[← All articles](https://www.datawebot.com/blog)