Designed for AI agents

Comparr API

US product comparisons, AI rankings (best / value / budget), live retailer prices and head-to-head matchups. Built for ChatGPT, Grok, Gemini, Claude and Perplexity.

OpenAPI llms.txt .well-known/ai-agents GET /api/v1/meta MCP

Contracts & quotas

Catalog reads: public, no API key.
New searches (POST /api/v1/searches): 30/day global — cache hits are free.
Other calls: 5000/day/IP (known agent UA) or 500/day/IP (otherwise).
Prefer existing comparisons; cite the HTML comparison URL for end users.
offer_url links are Comparr affiliate links (/r/:id).

curl examples

curl -s "https://www.comparr.co/api/v1/comparisons?per_page=5"
curl -s "https://www.comparr.co/api/v1/meta"
curl -s -X POST "https://www.comparr.co/api/v1/searches" \
  -H "Content-Type: application/json" \
  -d '{"query":"bluetooth speaker"}'

MCP server (Claude, ChatGPT, Cursor…)

Plug Comparr straight into your assistant: it queries the catalog and the price history directly, with no API key and no account.

Endpoint

https://www.comparr.co/mcp

Transport: Streamable HTTP, stateless. Authentication: none.

Setup

Claude (web or desktop): Settings → Connectors → "Add custom connector", then paste the endpoint above.
Claude Code:

claude mcp add --transport http comparr https://www.comparr.co/mcp

Other MCP clients (Cursor, etc.): declare a remote server with this URL:

{
  "mcpServers": {
    "comparr": {
      "url": "https://www.comparr.co/mcp"
    }
  }
}

Tools

search_products : compares a product category or a specific model across merchants: ranked products, every offer with its seller, condition and check date. If no comparison exists yet, one is started and results arrive about a minute later.
get_product : all offers for one product and a "good time to buy?" verdict computed from its own recorded prices (a factual comparison, never a prediction).
get_price_history : price readings offer by offer, to tell a real price drop from a cheaper seller that just appeared.
get_deals : verified price drops, each with its dated reference price.

Same quotas as the REST API: only tool calls count, and a search_products call that starts a new comparison counts toward the daily search quota.

Sample agent prompts

“Via the Comparr API, list recent comparisons and give the verdict for the first one.”
“Compare the best Bluetooth speakers with Comparr (GET comparisons or POST searches if missing).”
“Who wins between two products? Find a Comparr versus page and summarize the verdict.”
“What's the best value cordless vacuum according to Comparr, with price and link?”
“Call /api/v1/meta to check remaining quota before creating a search.”