Trueprice for AI assistants
Last updated November 19, 2026
The internet's best prices, found by AI. Trueprice's AI shopping agents find coupon codes, test each one in a real cart and track prices at 145,000+ stores. This server lets Claude, ChatGPT, Cursor and other AI assistants ask for both: a product's price history, whether now is a good price, and the codes that worked.
Connect
Remote server over Streamable HTTP. No sign-in, no API key.
https://checktrueprice.com/api/mcp/shopping- Claude: Settings → Connectors → Add custom connector, paste the URL.
- Cursor, VS Code, ChatGPT developer mode and other MCP clients: add a remote MCP server of type streamable-http with the URL.
Tools
All 8 tools read Trueprice's own data. Each one except cheapest_path is read-only; cheapest_path also saves its answer as a snapshot so every number can be reproduced.
lookup_price: True price (the lowest price the product really sold at that we know of, from daily checks and archived snapshots: `true_price`), price history, deal score, best checked promo code and product spec (description, tags, options, SKUs, weights, images) for a product page URL of a Shopify store Trueprice tracks (e.g.search_products: Find products by title words across every store Trueprice tracks.find_products: Filter every product Trueprice tracks across its Shopify stores, like the Product Finder at /product-finder (same filters, same results as GET /api/v1/finder).product_history: True price (`true_price`: lowest price it really sold at, from daily checks and archived snapshots), Stats per 30/90/180/365-day and all-time window (`stats.windows`: low/high with dates, average, price at window start, out-of-stock %, drop count, is_lowest, last_price_change), star rating and review count with history (`rating`), reviews from the store's review app (`reviews`: rating, count, star histogram, 3 recent reviews, Q&A, what reviewers mention), category rank among same-type products across stores (`category_rank`: #cheapest of N in stock, 30-day drop rank), units in stock per option (`stock`: qty, null = not limited, with its change history; `variants[].stock`), live price, every recorded price change, archived snapshots from before tracking began (`archived`, labelled "archived snapshot", not used in stats or score), 30/90-day and all-time low/average/high, deal score, best checked promo code and product spec (description, tags, options, SKUs, weights, images) for one product.lookup_gtin: Every tracked Shopify listing with this barcode (GTIN / UPC / EAN), across stores: domain, handle, title, price and stock of that option, and its Trueprice URL.store_codes: Promo codes for a store that Trueprice applied in a real cart and the store accepted.cheapest_path: The cheapest way to buy a product Trueprice tracks (same as POST /api/v1/prices/optimize): the exact steps (sale, each code, shipping, tax, rewards) with each step's amount, source_id (the code's deals id or the product URL), confidence and checked_at, then checkout_total = items − eligible discounts + shipping + tax (tax only with user.taxRate), effective_cost after rewards (null without one), codes rejected and why (not stackable, minimum not met, new customers only, expired), assumptions, and snapshot_id: the purchase_paths row that reproduces every number.find_store: Store domains for a brand or site name (e.g.
Try it
- “Is the price of this a good one? https://… (any Shopify product page)”
- “What promo codes work at this store right now? (a store name or domain)”
- “Find the cheapest way to buy this, with codes and shipping.”
- “Which products dropped 30% or more in the last 30 days and are in stock?”
For developers
https://checktrueprice.com/api/mcp serves the same tools plus 3 more for developers (the App Finder and TikTok Shop data). There is also a plain JSON API at https://checktrueprice.com/api/v1, described in llms.txt.
Limits and data
Free, with rate limits. Price history starts the day we first read a product (November 13, 2026 for the first stores); earlier prices appear only as labelled archived snapshots. Codes are the ones we tested, and a code that stops working comes off. How we check: how we verify.