Deal.ai's MCP server gives your editor's agent real store prices: every store's price for the same item, price history, live deals, deal checks and Deal.ai's shopping agent. Research parts for a build, check a purchase for the team, or put real prices into a shopping app prototype without leaving the editor.
Free with a Deal.ai account. No API key: you sign in once, in the browser.
Server address (Streamable HTTP, OAuth sign-in)
https://deal.ai/mcpCursor opens and asks to install the server. The first time it connects, Deal.ai's sign-in opens in your browser; sign in and choose Allow.
Add this to ~/.cursor/mcp.json for every project, or to .cursor/mcp.json in a project's folder for that project only:
{
"mcpServers": {
"deal-ai": { "url": "https://deal.ai/mcp" }
}
}Turn the server on or off from Customize in Cursor's sidebar (older versions list MCP servers in Cursor's settings). The agent asks before it runs a tool unless your approval settings say otherwise.
Deal.ai has also been submitted to the Cursor Marketplace as a plugin, and is waiting for review. Until it's listed, the button and the mcp.json entry above are the way in; the plugin connects to the same server.
VS Code shows the server's details to review before you install it. Its sign-in opens in your browser on the first connection.
Run MCP: Add Server from the Command Palette, choose HTTP and paste the server address above. Or add it to a workspace's .vscode/mcp.json:
{
"servers": {
"deal-ai": { "type": "http", "url": "https://deal.ai/mcp" }
}
}Then ask in Copilot Chat. Copilot picks Deal.ai's tools when a question needs them, and asks before it runs one.
Other editors and agents work too: any client that supports remote MCP servers over Streamable HTTP, with OAuth sign-in. The server's metadata and a ready .mcp.json are in our public repository, Famous-Labs/deal-ai-mcp. Disconnect an editor any time from Connected apps on your account page; it lapses by itself after 90 days unused.
Name Deal.ai in the request so the agent reaches for it. Every tool returns structured data beside its text, and every answer links to the Deal.ai page behind it.
“I'm prototyping a price-tracker app. Get every store's current price for the Sony WH-1000XM5 from Deal.ai and write a TypeScript type that matches the result.”
Answered by search_products and get_product_prices
Structured results as well as text. Each store's price comes with store, price, shipping, condition, availability, price_seen_at, buy_url and a best flag; the price history has lowest, highest, stores_tracked and recent_daily_lows. Enough to shape a type and a first screen around real data.
“Research parts for a home server build: compare the Samsung 990 Pro and the WD Black SN850X, 2TB, on Deal.ai.”
Answered by search_products and compare_products
The products side by side: the specs they share, row by row, each one's rating, and its best price from a store with when it was seen and a link, plus Deal.ai's comparison page for the pair when it has one.
“Ask Deal.ai for a 27-inch 4K monitor under $350 with USB-C power delivery, for coding on a MacBook.”
Answered by ask_deal_ai and get_deal_ai_answer
Deal.ai's shopping agent researches it in one to three minutes, and the editor's agent checks back. You get its answer and up to six picks, each with why it fits, its best store price and when that was seen, its rating and links. Follow-ups (“only matte screens”) continue the same search.
“Before I order this for the office, check it with Deal.ai: (paste a link to the product at any store)”
Answered by check_deal and get_deal_check
A deal check of the exact item, new, at every big US store at the same moment, usually in 20 to 70 seconds: a verdict (cheaper elsewhere, as low as any store, or the only store), each store's price with a link, whether the “was” price holds up, and a result page to share with whoever approves the order.
“Look up the Raspberry Pi 5 8GB on Deal.ai: specs, what's in the box and the best price.”
Answered by search_products and get_product_details
Its description and specs, a fact sheet read from a store's own page (highlights, what's in the box, returns, what buyers praise and complain about), its rating, its photos, and its best price from a store with a link.
“For a demo of my deals widget, pull today's Deal.ai tech deals over 25% off and save them to fixtures/deals.json.”
Answered by get_deals
Live deals in that category, each with its title, price, the price it's measured against and what that is, the discount, the store, when it was last checked, and links to its deal page and the store. The prices are real and change, so the file is a snapshot.