Findcheap API

Authentication

Every request you send must carry an API key, which you can create from the developer dashboard. Keys begin with fc_live_….

Send the key in the Authorization header — a header is a line of metadata that rides along with an HTTP request, separate from its body:

Authorization: Bearer fc_live_xxxxxxxxxxxxxxxx

For quick tests, a token query parameter is also accepted (?token=fc_live_…). Use the header in anything you ship — query strings end up in server logs.

Quickstart

Every Findcheap call has the same simple shape: your code sends an HTTPS POST request (the same mechanism a browser uses to submit a form) to an endpoint URL, carrying your API key in a header and a small JSON body that says what to search. Findcheap answers with JSON.

Here’s a complete first call — finding cheaper alternatives from nothing but a product page URL:

# One complete search. -H adds a header (your key,
# then the body's format); -d is the JSON body.
curl https://api.findcheap.ai/v1/search_url \
  -H "Authorization: Bearer $FINDCHEAP_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://www.amazon.com/dp/B0BTYCRJSS"}'

That’s the whole integration — there is no step two. A typical response (trimmed to two results):

{
  "results": [
    {
      "title": "Stanley Quencher H2.0 FlowState Tumbler, 40 oz",
      "price": 27.99,
      "currency": "USD",
      "url": "https://www.walmart.com/ip/...",
      "image": "https://i5.walmartimages.com/...",
      "marketplace": "Walmart",
      "rating": 4.8,
      "ratings_count": 11204
    },
    {
      "title": "Stanley 40oz Quencher H2.0 Tumbler",
      "price": 29.50,
      "currency": "USD",
      "url": "https://www.ebay.com/itm/...",
      "image": "https://i.ebayimg.com/...",
      "marketplace": "eBay",
      "rating": null,
      "ratings_count": null
    }
  ],
  "result_count": 10
}

Search

Finds matching products with lower prices from structured product data — a title and image you already have (for example, from a page your agent is browsing, or a prior Extract call). Only have a URL? Use Search URL.

Request body

Response

Up to 10 matching products (a custom count is available on Enterprise), ordered by deal quality:

Search URL

Finds matching products with lower prices from nothing but a product page URL. Under the hood this chains the other two endpoints: Extract pulls the product’s title, image, price, and brand from the page, then Search runs on that data. One call, one response. Billed the same as 1 /extract and 1 /search call.

Request body

Response

Identical to Search — up to 10 matching products, ordered by deal quality.

Extract

Extracts structured product data from any product page URL — the same extraction stage that powers Search, exposed standalone. Useful when you want the product’s identity without running a full search.

Request body

Response

{
  "is_product_page": true,
  "title": "Stanley Quencher H2.0 FlowState Tumbler, 40 oz",
  "price": 35.00,
  "currency": "USD",
  "image": "https://m.media-amazon.com/images/I/...",
  "brand": "Stanley",
  "in_stock": true
}

When the URL isn’t a product page (a homepage, a category listing, an article), is_product_page is false and the remaining fields are null. price can be null on genuine product pages too — treat title and image as the reliable core. price is the page’s standing purchase price: the price any shopper pays, not first-order, coupon, or membership prices.

in_stock is true when the page says the product is currently purchasable, false when it’s out of stock or discontinued, and null when availability couldn’t be determined from the page.

Errors

Errors use conventional HTTP status codes with a JSON body:

{
  "error": {
    "type": "invalid_request",
    "message": "title and image_url are required."
  }
}

Failed requests are never billed.

MCP server

Findcheap runs a hosted Model Context Protocol server, so agents can use Findcheap without any integration code. It speaks Streamable HTTP and authenticates with the same API keys as the REST API:

https://mcp.findcheap.ai/mcp

MCP tool calls are billed identically to the endpoint they wrap — there’s no separate MCP pricing.

Tools

Client setup

Any MCP client that supports remote servers works. Two common setups:

// Works with Claude Code, Cursor, and any client that reads .mcp.json
{
  "mcpServers": {
    "findcheap": {
      "type": "http",
      "url": "https://mcp.findcheap.ai/mcp",
      "headers": {
        "Authorization": "Bearer fc_live_..."
      }
    }
  }
}

Once connected, the agent sees both tools and their schemas automatically. A prompt like “find this tumbler cheaper: <url>” is enough — the agent calls find_cheaper_products and works with the returned listings.